[House Hearing, 110 Congress]
[From the U.S. Government Publishing Office]


 
                 THE USE OF ARTIFICIAL INTELLIGENCE TO 
                IMPROVE THE U.S. DEPARTMENT OF VETERANS 
                   AFFAIRS' CLAIMS PROCESSING SYSTEM 

=======================================================================

                                HEARING

                               before the

       SUBCOMMITTEE ON DISABILITY ASSISTANCE AND MEMORIAL AFFAIRS

                                 of the

                     COMMITTEE ON VETERANS' AFFAIRS
                     U.S. HOUSE OF REPRESENTATIVES

                       ONE HUNDRED TENTH CONGRESS

                             SECOND SESSION

                               __________

                            JANUARY 29, 2008

                               __________

                           Serial No. 110-66

                               __________

       Printed for the use of the Committee on Veterans' Affairs

                               ----------
                         U.S. GOVERNMENT PRINTING OFFICE 

41-366 PDF                       WASHINGTON : 2008 

For sale by the Superintendent of Documents, U.S. Government Printing 
Office Internet: bookstore.gpo.gov Phone: toll free (866) 512-1800; 
DC area (202) 512-1800 Fax: (202) 512-2104 Mail: Stop IDCC, 
Washington, DC 20402-0001 




































                     COMMITTEE ON VETERANS' AFFAIRS

                    BOB FILNER, California, Chairman

CORRINE BROWN, Florida               STEVE BUYER, Indiana, Ranking
VIC SNYDER, Arkansas                 CLIFF STEARNS, Florida
MICHAEL H. MICHAUD, Maine            JERRY MORAN, Kansas
STEPHANIE HERSETH SANDLIN, South     RICHARD H. BAKER, Louisiana
Dakota                               HENRY E. BROWN, Jr., South 
HARRY E. MITCHELL, Arizona           Carolina
JOHN J. HALL, New York               JEFF MILLER, Florida
PHIL HARE, Illinois                  JOHN BOOZMAN, Arkansas
MICHAEL F. DOYLE, Pennsylvania       GINNY BROWN-WAITE, Florida
SHELLEY BERKLEY, Nevada              MICHAEL R. TURNER, Ohio
JOHN T. SALAZAR, Colorado            BRIAN P. BILBRAY, California
CIRO D. RODRIGUEZ, Texas             DOUG LAMBORN, Colorado
JOE DONNELLY, Indiana                GUS M. BILIRAKIS, Florida
JERRY McNERNEY, California           VERN BUCHANAN, Florida
ZACHARY T. SPACE, Ohio
TIMOTHY J. WALZ, Minnesota

                   Malcom A. Shorter, Staff Director

                                 ______

       SUBCOMMITTEE ON DISABILITY ASSISTANCE AND MEMORIAL AFFAIRS

                    JOHN J. HALL, New York, Chairman

CIRO D. RODRIGUEZ, Texas             DOUG LAMBORN, Colorado, Ranking
PHIL HARE, Illinois                  MICHAEL R. TURNER, Ohio
SHELLEY BERKLEY, Nevada              GUS M. BILIRAKIS, Florida

Pursuant to clause 2(e)(4) of Rule XI of the Rules of the House, public 
hearing records of the Committee on Veterans' Affairs are also 
published in electronic form. The printed hearing record remains the 
official version. Because electronic submissions are used to prepare 
both printed and electronic versions of the hearing record, the process 
of converting between various electronic formats may introduce 
unintentional errors or omissions. Such occurrences are inherent in the 
current publication process and should diminish as the process is 
further refined.
























                            C O N T E N T S

                               __________

                            January 29, 2008

                                                                   Page
The Use of Artificial Intelligence to Improve the U.S. Department 
  of Veterans Affairs' Claims Processing System..................     1

                           OPENING STATEMENTS

Chairman John J. Hall............................................     1
    Prepared statement of Chairman Hall..........................    44
Hon. Doug Lamborn, Ranking Republican Member.....................     4
    Prepared statement of Congressman Lamborn....................    45

                               WITNESSES

U.S. Department of Veterans Affairs:
  Kim A. Graves, Director, Office of Business Process 
    Integration, Veterans Benefits Administration................    33
      Prepared statement of Ms. Graves...........................    78
  Stephen W. Warren, Principal Deputy Assistant Secretary for 
    Information and Technology, Office of Information and 
    Technology...................................................    35
      Prepared statement of Mr. Warren...........................    79

                                 ______

Christopherson, Gary A., University Park, MD (Former Senior 
  Advisor to the Under Secretary for Health, and Chief 
  Information Officer, Veterans Health Administration, U.S. 
  Department of Veterans Affairs, Former Principal Deputy 
  Assistant Secretary for Health Affairs, U.S. Department of 
  Defense).......................................................    24
    Prepared statement of Mr. Christopherson.....................    75
Cleveland, Gunnery Sergeant Tai, USMC (Ret.), Dumfries, VA.......     5
    Prepared statement of Gunnery Sergeant Cleveland.............    45
Miller, Randolph A., M.D., Donald A.B. and Mary M. Lindberg 
  University Professor of Biomedical Informatics, Medicine, and 
  Nursing, Vanderbilt University School of Medicine, Nashville, 
  TN.............................................................    18
    Prepared statement of Dr. Miller.............................    53
Mitchell, Tom M., Ph.D., E. Fredkin Professor and Chair, Machine 
  Learning Department, School of Computer Science, Carnegie 
  Mellon University, Pittsburgh, PA..............................    16
    Prepared statement of Mr. Mitchell...........................    49
QTC Management, Inc., Diamond Bar, CA, Marjie Shahani, M.D, 
  Senior Vice President, Operations..............................    19
    Prepared statement of Dr. Shahani............................    58
Stratizon Corporation, Atlanta, GA, Ned M. Hunter, President and 
  Chief Executive Officer........................................    21
    Prepared statement of Mr. Hunter.............................    61
Unum, Portland, ME, John F. McGarry, Senior Vice President of 
  Benefits, Chief Risk Officer...................................    23
    Prepared statement of Mr. McGarry............................    64
Wounded Warrior Project, John Roberts, National Service Director.     8
    Prepared statement of Mr. Roberts............................    47

                       SUBMISSIONS FOR THE RECORD

American Veterans (AMVETS), Raymond C. Kelley, National 
  Legislative Director, statement................................    80
American Legion, Steve Smithson, Deputy Director, Veterans 
  Affairs and Rehabilitation Commission, statement...............    82
Disabled American Veterans, Kerry Baker, Associate National 
  Legislative Director, statement................................    82
Paralyzed Veterans of America, statement.........................    83

                   MATERIAL SUBMITTED FOR THE RECORD

Post Hearing Questions and Responses for the Record:
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Tom Mitchell, Ph.D., School of Computer Science, 
  Machine Learning Department, Carnegie Mellon University, 
  Pittsburgh, PA, letter dated February 4, 2008, and April 2008 
  response from Mr. Mitchell.....................................    85
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Randolph Miller, Ph.D., Professor and Former Chair, 
  Department of Biomedical Informatics, Vanderbilt University 
  School of Medicine, Nashville, TN, letter dated February 4, 
  2008, and February 28, 2008, response from Dr. Miller..........    87
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Marjie Shahani, M.D., QTC Management, Inc., Diamond 
  Bar, CA, letter dated February 4, 2008, and response letter 
  dated March 3, 2008............................................    95
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Ned M. Hunter, President and Chief Executive 
  Officer, Stratizon Corporation, Atlanta, GA, letter dated 
  February 4, 2008, and March 6, 2008, response from Mr. Hunter..    97
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Gary A. Christopherson, Strategic Management and 
  Performance, University Park, MD, letter dated February 4, 
  2008, and response from Mr. Christopherson.....................    98
Hon. John J. Hall, Chairman, Subcommittee on Disability 
  Assistance and Memorial Affairs, Committee on Veterans' 
  Affairs, to Hon. James B. Peake, M.D., Secretary, U.S. 
  Department of Veterans Affairs, letter dated February 4, 2008, 
  and VA responses...............................................   100


                 THE USE OF ARTIFICIAL INTELLIGENCE TO
                     IMPROVE THE U.S. DEPARTMENT OF
                        VETERANS AFFAIRS' CLAIMS
                           PROCESSING SYSTEM

                              ----------                              


                       TUESDAY, JANUARY 29, 2008

             U.S. House of Representatives,
                    Committee on Veterans' Affairs,
Subcommittee on Disability Assistance and Memorial Affairs,
                                            Washington, DC.

    The Subcommittee met, pursuant to notice, at 2:06 p.m., in 
Room 340, Cannon House Office Building, Hon. John J. Hall 
[Chairman of the Subcommittee] presiding.
    Present: Representatives Hall, Lamborn, Bilirakis.

               OPENING STATEMENT OF CHAIRMAN HALL

    Mr. Hall. Good afternoon ladies and gentlemen. The 
Committee on Veterans' Affairs, Subcommittee Disability 
Assistance and Memorial Affairs, hearing on ``The Use of 
Artificial Intelligence to Improve the U.S. Department of 
Veterans Affairs' (VA's) Claims Processing System'' will come 
to order.
    Before I begin with my opening statement, I would like to 
call attention to the fact that Raymond C. Kelley, National 
Legislative Director for AMVETS and Kerry Baker, Associate 
National Legislative Director for the Disabled American 
Veterans have asked to submit written statements for the 
hearing record.
    If there is no objection, I ask unanimous consent that 
these statements be entered into the record.
    Hearing no objection, so entered.
    [The statements of Mr. Kelly and Mr. Baker appear on p. 80 
and p. 82.]
    I would ask that we all rise for the Pledge of Allegiance. 
The flag is in this corner of the room.
    [Pledge of Allegiance.]
    I would like, first of all, to thank the witnesses for 
coming today to appear before the Subcommittee. I know I speak 
for my colleagues when I say we are all extremely frustrated 
and disappointed when we hear about 650,000 claims pending and 
another 147,000 appeals with a delay of 183 days to process 
those claims.
    But looking at this photograph, which is up on the screen 
right now, of an eight-inch paper record held together with 
rubber bands and marked with post-it notes, it is hard to 
imagine that things do not get lost or missed. This has got to 
be cumbersome when processing our veterans' claims.

[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]


    There is no doubt that we need a better system than rubber 
bands and post-it notes and must look beyond the current way VA 
is doing business. There are best practices within the 
scientific community and best practices in use in the private 
sector.
    I thank you for joining me and the Subcommittee today to 
explore these solutions to broaden our understanding of what is 
possible, realistic, and achievable in this technological age.
    The current VA claims process is paper intensive, complex 
to manage, difficult to understand, and takes years to learn. 
Training a rater can take 2-3 years and many leave within 5 
years. Experienced raters can adjudicate about 3 claims a day, 
taking about 2-3 hours apiece.
    This means that if there are 10 people who can rate a claim 
and 800 claims are ready to rate, then it will take another 80 
days to process those pending claims, which have already been 
in the system for several months.
    This is very labor intensive. And in the meantime, veterans 
are waiting months without compensation while their completed 
case sits on a shelf. I know the other Members of the Committee 
and most Americans find that unacceptable.
    Additionally, there have been reports by the U.S. 
Government Accountability Office, the VA Inspector General, and 
the Institute for Defense Analyses that explored the variances 
in ratings between Regional Offices (ROs) and the lack of 
inter-rater reliability.
    The Veterans' Disability Benefits Commission also found a 
great deal of subjectivity and inconsistency in the VA's 
disability claims process.
    So how do we solve this?
    I have had a lifelong interest in science, was a three-time 
National Science Foundation scholar, and a physics student 
while at Notre Dame. I learned FORTRAN when I was a kid when my 
father was teaching seminars when the computer would take up a 
room this size that now fits into a laptop.
    So I find the topic of artificial intelligence, or AI, 
compelling since it requires the confluence of science, 
technology, mathematics, engineering, and physics.
    In general, the purpose of AI is to make computer programs 
or machines that can solve problems and achieve goals. AI 
software increases speed, improves accuracy and reduces costs 
for many industries and agencies.
    AI does not replace the human element, but rather 
facilitates its availability. There are many examples of AI in 
other areas, such as banking and medicine. For instance, the 
Veterans Health Administration (VHA) relies on VistA to help 
doctors with diagnosis and treatment. It sends alerts when a 
patient needs a flu shot, cholesterol screening, or warns of 
potential drug interactions.
    AI can be a decision support tool for adjudicating claims 
too. It could be used to organize and store data. It could 
match key words from a veteran's record to the criteria in the 
Rating Schedule. It could prioritize multiple disability 
issues.
    I envision a VA in which a veteran can apply online for 
benefits, upload records, exams, and other certificates, which 
are prioritized and classified by an expert system that can 
match the data to the Rating Schedule criteria and thereby 
shorten the time it takes to generate a claim.
    The electronic template used by the examiner could be 
associated with the Rating Schedule, which could also help 
calculate ratings. Classifiers or key words could easily be 
matched by the computer to the Rating Schedule, such as 
``Arm,'' ``Amputation,'' then ``90 percent.''
    This would free up the time for the RO employers to deal 
with the more complicated issues, and assist veterans and their 
families with their problems.
    This Subcommittee has often heard that veterans do not know 
about, or understand, their benefits and that transitioning 
servicemembers are not getting all of the support that they 
need from the Veterans Benefits Administration (VBA).
    In this way, VBA staff could be providing more outreach and 
ensuring that veterans understand their entitlements and 
eligibility requirements for other programs, and benefits such 
as vocational rehabilitation, insurance and special monthly 
compensation.
    I am eager to hear testimony today that will open up the 
discussion on information technology (IT) and share ideas that 
can improve rating efficiency, quality, and accuracy while 
reducing inconsistencies and variances in decisions for our 
disabled veterans who often have been waiting for a long time 
for a claim determination.
    I look forward to working with Ranking Member Lamborn and 
the Members of this Subcommittee in finding real solutions that 
will vastly improve the VA claims process. It is unconscionable 
that our veterans are waiting as long as they are for their 
earned benefits. And this situation must end.
    I now recognize the distinguished Mr. Lamborn for any 
opening remarks he may have.
    [The prepared statement of Chairman Hall appears on p. 44.]

             OPENING STATEMENT OF HON. DOUG LAMBORN

    Mr. Lamborn. Thank you, Mr. Chairman for yielding. I would 
like to welcome all of our witnesses to this Subcommittee's 
first hearing of the Second Session.
    I want to commend you, Mr. Chairman, for your leadership 
and bipartisanship in the previous session. And I look forward 
to working with you and your staff to find meaningful solutions 
to improving the VBA claims processing system and reducing 
VBA's disability claims backlog.
    I am excited that our topic of discussion today is the Use 
of Artificial Intelligence to Improve the Disability Claims 
Process.
    As you know, Mr. Chairman, this is an idea that my 
colleagues and I on this side of the aisle have long supported.
    Whether it was in our fiscal year 2008 views and estimates, 
or two bills that I introduced last session, H.R. 1864 and H.R. 
3047, we believe that one way to truly reduce the current 
backlog and prevent future backlogs is to propel the VA beyond 
a 20th century, paper-based processing system, as you so 
eloquently showed us through a picture on the screen.
    VA must create a system where all claims are electronically 
scanned and rating board members have access to computerized 
interactive tools to assist them in the adjudicative process.
    Hopefully, the new system will lead to more accurate rating 
decisions that are delivered to our Nation's veterans in a 
timely manner.
    While I envision an important role for artificial 
intelligence in the decisionmaking process, I also concur with 
our witnesses who will attest that this technology should not 
and will not ever completely replace claims adjudicators.
    A few weeks ago, staff from both sides of the aisle 
attended a briefing where VBA laid out plans to move forward 
with such a system. And I am excited to learn more about those 
plans today.
    The Subcommittee must ensure that this new initiative is 
fully funded and completed with the speed and attentiveness 
that our veterans deserve.
    I am glad that we have representatives from both the 
private and academic sectors here with us today. It is my hope 
that they will be able to help VA develop some of the options 
that are currently available in the private sector.
    While I understand that VA has a very large and unique 
disability claims system, there are similar systems out there. 
And I would hope that VA would look at these systems before 
they reinvent the wheel.
    We must improve this system so heroes like Gunnery Sergeant 
Cleveland do not have to wait several years to have their claim 
adjudicated correctly.
    Mr. Chairman, I extend my thanks to you and your staff for 
holding this hearing this afternoon. And I look forward to 
hearing the testimony of our witnesses. I yield back the 
balance of my time.
    [The prepared statement of Congressman Lamborn appears on 
p. 45.]
    Mr. Hall. Thank you, Congressman Lamborn.
    Joining us on our first panel is Tai Cleveland from 
Dumfries, Virginia. Mr. Cleveland is a medically-retired Marine 
who sustained a devastating training injury in Kuwait in 2003. 
With him is his wife, Robin.
    And they are joined by John Roberts, the National Service 
Director for Wounded Warrior Project (WWP), which is the 
veterans service organization (VSO) that represents the 
Clevelands. I would also like to recognize Mr. Roberts' 
distinguished service as a Marine who was also severely injured 
while serving this Nation in Somalia. I thank you all for being 
here.
    I would like to remind our panelists that your complete 
written statements have been made a part of the hearing record.
    Therefore, if you would, try to limit your remarks to 5 
minutes so that we have sufficient time for follow-up 
questions.
    Mr. Cleveland, we will go ahead and begin with your 
testimony. You are now recognized, sir.

  STATEMENTS OF GUNNERY SERGEANT TAI CLEVELAND, USMC (RET.), 
     DUMFRIES, VA (DISABLED VETERAN); ACCOMPANIED BY ROBIN 
  CLEVELAND, DUMFRIES, VA; AND JOHN ROBERTS, NATIONAL SERVICE 
               DIRECTOR, WOUNDED WARRIOR PROJECT

    STATEMENT OF GUNNERY SERGEANT TAI CLEVELAND, USMC (RET.)

    Mr. Cleveland. Mr. Chairman, Ranking Member Lamborn, 
distinguished Members of the Committee, thank you for the 
opportunity to testify before you regarding my experience with 
the Department of Veterans Affairs and claims process. My name 
is Gunnery Sergeant Tai Cleveland, United States Marine Corps 
Retired. With me today I have my wife, Robin. And I would like, 
with your permission, Mr. Chairman, for my wife, who has dealt 
often with the VA on our benefits claim, to discuss the issues.
    Mr. Hall. Thank you, sir. And, Robin, you are now 
recognized for 5 minutes.
    Mrs. Cleveland. Thank you, Mr. Chairman. My husband served 
his country proudly for 24 years as a United States Marine. And 
although we had many issues with the U.S. Department of Defense 
(DoD) following his injuries, due to the subject of this 
hearing, I will limit my comments to our difficulties with the 
VA claims processing system and its impact on our family. As I 
am speaking, however, please keep in mind that a severely 
injured servicemember must navigate multiple systems: the 
Department of Defense, the Social Security Administration, 
Medicare, and the VA. It is quite overwhelming to say the 
least.
    Tai was injured in August 2003 during a hand-to-hand combat 
training accident in Kuwait, where he was flipped onto his 
back, injuring his head and multiple vertebras. The resulting 
damage has left my husband a paraplegic with chronic 
neuropathic pain, spasticity, and what is classified as a mild-
to-moderate traumatic brain injury that has its own set of 
challenges.
    Since Tai's injury, I have had to learn the hard way how to 
navigate the systems. Keeping meticulous records of documents, 
recording dates and times of telephone calls, confirming 
receipt of anything sent or hand delivered to Federal agencies.
    As such, I thought the best way to convey our situation was 
to share a timeline detailing our experiences with the VA.
    In June 2005, we attended the Transition Assistance Program 
class provided by the Marine Corps and the VA to learn about 
the available options. We completed the VA's Benefits Delivery 
at Discharge (BDD) process, including the benefits, specially 
adaptive housing, and adaptive vehicle program applications, 
and hand delivered it with medical records, MRI compact discs, 
films, prescription reports, et cetera, in its totality to 1722 
Eye Street, Washington, DC.
    After having completed his compensation and pension exam, 
we called the VA Benefits number in November of 2005 where we 
were advised that the application was incomplete and medical 
records from the military treatment facility (MTF) were needed. 
I delivered a second copy of MTF medical records to the DC 
Office.
    A month later, I phoned again to see if the records were 
received and was advised that no application was on file. I 
copied and redelivered the original application to the DC 
Office.
    In January 2006, another call to VA Benefits advised me 
that the claim was being reviewed, but that medical records 
were required to make a final determination.
    I again copied medical records and redelivered to the DC 
Office. I was later told that the housing and vehicle grant 
were denied.
    When I called in February of 2006, I was told no 
determination could be made because Tai was still on active 
duty. Additionally, I was told that no claim was on file for 
the housing or vehicle. We reapplied.
    In March 2006, I met with a VA employee at Walter Reed 
regarding benefits and our difficulty with the claims. She 
introduced us to a VA social worker at Walter Reed who enrolled 
Tai in the Adaptive Driving Program at Richmond.
    We were told to reapply for benefits, because no 
application was found. We resubmitted the original application 
and completed a new application for Specially Adaptive Housing, 
Home Improvement and Structural Alteration (HISA), and the 
vehicle grant, but were informed on April 5th that the 
applications were denied and advised to reapply.
    In June 2006, we were informed by the VA social worker that 
the approval for the vehicle application was in fact received. 
But she was ``unable to locate the application, because the 
clerk failed to separate the application and maintain an in-
house copy.''
    In addition, our HISA and Adaptive Housing Grants were 
denied. We reapplied.
    Everything was quiet for the next 3 months until October 
26, 2006, when we were advised to reapply for vehicle and 
housing grants since no official notification of approval was 
received.
    Again, in November of 2006, we received verbal notification 
from the VA representative at Walter Reed of the latest vehicle 
and housing denial. And on December 13th, 2006, we were advised 
to reapply for vehicle and housing grants, and were contacted 
by VA to verify our address.
    In January of 2007, Tai was medically retired from the 
Marine Corps. After filing BDD, we assumed we would get his 
disability check within a month or so.
    In February of 2007, our housing and vehicle grants were 
approved and supposedly had been approved since April of 2006. 
But the hard copy was no longer on file. To date, we still have 
not received the official vehicle approval.
    In late May 2007, we received verbal notification from the 
VSO, helping us at the time, that the VA was indicating that 
there was not enough information on file to rate the claim. 
And, therefore, additional information was necessary.
    In June, we received notification from the VA of an 80 
percent partial rating. We were advised that the rating was 
temporary and additional information was necessary in order to 
process the claim.
    As we were scheduled to be in Richmond shortly to obtain an 
adaptive cycle, we were advised to have Richmond perform the 
necessary evaluations for submittal to the Roanoke Regional 
Office.
    While at Richmond, I also inquired about obtaining the 
vehicle grant hard copy and contacted the VA to inquire about 
Aid and Attendance. I was told that I was not eligible.
    In July 2007, via express mail, Tai's medical records from 
Richmond to Roanoke--I delivered Tai's records from Richmond to 
Roanoke and sent the VA an email advising that we still had not 
received a disability check approximately 6 months post-
discharge.
    In August, I phoned and emailed VA Benefits again and told 
them that despite the temporary rating, we still had no check. 
I requested direct deposit information and requested to verify 
our address.
    After having been contacted about our problems by a non-
profit organization, a concerned representative from the VA's 
Central Office called in September about the outstanding 
checks. And we were told that a tracer would have to be placed 
on the missing checks before replacements could be mailed. I 
later received a call from the Roanoke office and was advised 
that replacement checks were going to be issued.
    On October 4th, 2007, a VA representative told us that the 
claim was being expedited and should be completed by the 14th. 
We were informed on the 14th and on the 30th that the updated 
medical reports still had not been received. However, on the 
29th we began to receive the replacement checks for the 
temporary rating.
    At this point in the timeline, it is important to note that 
our family had now been without our full disability 
compensation and benefits for almost 11 months. Our college-
aged children were forced to withdraw. The overall financial 
strain, and frustration level, and emotional toll, in addition 
to the actual injury, were crushing.
    Finally, on January 7th, 2008, after the intervention of 
Mr. Hall's Subcommittee and the Wounded Warrior Project, we 
received a final rating and back payment totaling thousands of 
dollars.
    As you can see we filed and re-filed, submitted and 
resubmitted medical records, claims forms, applications, and so 
on. But no one seemed to be able to track anything, placing 
additional burdens on an already overwhelmed family. In our 
case, after the intervention of a Congressional office and a 
non-profit organization, we were able to get the benefits Tai 
has earned. This process should not be that hard.
    Today, almost 4 years later, while we still have a few 
things to resolve with our ratings and benefits, our family is 
trying to move on.
    Many people have stepped in to help us, from government 
agencies, to Congressional offices, to non-profit 
organizations. I am planning to return to work and school. Our 
children are returning to school. And Tai is enrolled in a 
media careers program for veterans in Chairman Filner's 
district. He has been a noted leader in the program, and ever 
the Gunny, and has even spoken to the Wounded Warrior Project 
about being a peer mentor.
    However, our purpose in coming here today is not only to 
tell you our story, but also to let you know that we are not 
alone. People we know have had similar problems. And we know 
there are more out there.
    We are hoping that our presence here will help you 
understand the obstacles faced by wounded members and their 
families and inspire everyone involved to work together to 
improve the efficiency of this vital system for the benefit of 
those who sacrificed so much for this country.
    Thank you, and I look forward to any questions you may 
have.
    [The prepared statement of Gunnery Sergeant Cleveland 
appears on p. 45.]
    Mr. Hall. Thank you, Mrs. Cleveland, and thank you so much 
Gunnery Sergeant Cleveland for your testimony. And--If your 
case was expedited, I would hate to see one that was not 
expedited.
    Now we recognize John Roberts from the Wounded Warrior 
Project. Mr. Roberts, you are recognized for 5 minutes.

                   STATEMENT OF JOHN ROBERTS

    Mr. Roberts. Mr. Chairman, Ranking Member Lamborn, 
distinguished Members of the Subcommittee, thank you for the 
opportunity to testify before you today regarding the use of 
technology to improve the efficiency of the Department of 
Veterans Affairs' claims process.
    My name is John Roberts. And I am the National Service 
Director for the Wounded Warrior Project, a non-profit, non-
partisan organization dedicated to assisting the men and women 
of the United States Armed Forces who have been injured during 
the current conflicts around the world.
    As a result of our direct, daily contact with these wounded 
warriors, we have a unique perspective on their needs and the 
obstacles they face as they attempt to transition and 
reintegrate into their communities.
    In addition to my experience with the Wounded Warrior 
Project in general and the Clevelands' case specifically, I am 
a service-connected veteran, a former veteran service officer, 
and most recently a supervisor with the Houston VA Regional 
Office where I had the opportunity to review claims and became 
familiar with a number of significant deficiencies within the 
system.
    In order to fully appreciate the problem, it is important 
to understand how the systems currently operate. Despite recent 
advances in technology common to most businesses, the Veterans 
Benefits Administration claims processing system is still 
dependent on a paper system. Although the VBA can now view 
electronic health records transmitted from the Veterans Health 
Administration, the ratings team is still required to print the 
records, place them in the veteran's claim folder, which are 
then reviewed page by page by a Veteran or a Rating Veteran 
Service Representative (RVSR).
    The current model of the VBA claims processing system has a 
total of six separate teams and often, but not always, includes 
another team that is dedicated to the processing of the 
Operation Iraqi Freedom/Operation Enduring Freedom (OIF/OEF) 
cases.
    The six main teams are, of course, triage, which handles 
the incoming claims, evidence, and is charged with maintaining 
the outdated file cabinet system, which stores the hard copy 
paper claims files.
    Predetermination, is charged with the initial development 
of all claims for Service-connected disabilities.
    The rating team is responsible for reviewing all available 
evidence and determining if the disabilities are service 
related. If so, they also assign the disability percentage.
    The post-determination team is responsible for inputting 
awards and generating notification letters to the claimants.
    The appeals team maintains all pending appeals submitted by 
all claimants.
    And the public contact team is charged with the general 
phone calls, questions, and conducting one-on-one interviews 
with the veterans, dependents, and survivors.
    Files must be hand carried to each of the teams. And any 
member of these teams has access to the records at any given 
time.
    Despite the number of people with access and the ease with 
which files may be misplaced, VBA only has one way to locate 
the files once it is removed from the filing cabinet.
    An electronic system called COVERS, but this system is only 
effective if utilized by the individual employee. Rather than 
having access to the file through electronic means, COVERS 
requires manual input to identify a specific location or 
individual. If this is not done, it is very time consuming to 
locate one file among all the files that are within the 
processing system.
    I'll give you an example. Within the Houston Regional 
Office, there are approximately 200 employees. And each of 
these employees could have up to 30 or more files at his or her 
desk at any time.
    Another challenge is the outdated filing system, which is 
used to store thousands of active files warehoused either at or 
near Regional Offices. If a file clerk or an employee for that 
matter is not paying attention and misfiles a claim folder into 
the wrong cabinet or drawer, it then becomes a very time 
consuming and difficult task to check each and every drawer to 
locate the missing file.
    The Triage Team at each RO is responsible for the intake of 
all new claims and evidence submitted by each and every 
claimant. If the file is not easily located, the mail is placed 
on search within the COVERS system until the file can be 
located.
    Because there are so many teams within the claims 
processing system, a particular file could be located within 
teams at any given time. This allows for--this allows for the 
human error factor, which is often why the numerous pieces of 
vital evidence are often lost or misplaced and cannot be 
associated with the appropriate claim folder.
    If a file cannot be located and all avenues have been 
exhausted to locate the file, the Regional Office will take 
action to rebuild the folder from scratch. This means that all 
prior evidence, claims, and claims which are submitted by the 
claimant are then lost. The responsibility to replace the 
missing evidence or claims is placed on the claimant. The VA 
will ask the claimant to submit any copies that he or she may 
have in their possession.
    In addition, due to the current war on terrorism, VBA is 
faced with another challenge. The new challenge is trying to 
obtain records from the National Guard and Reserve units. 
Active duty forces obviously do not file a claim until released 
from service. Once demobilized, a Reserve member or National 
Guard component is eligible to file such a claim. If 
reactivated, however, the Reservist's claim is halted and he or 
she at that time will take their service medical records with 
them into theater.
    There is also the large backlog of records requests to the 
Records Management Center, which houses not only claim folders, 
but now receives all servicemember records for recently 
discharged servicemen. Think of this as a large warehouse of 
nothing but paper files and an inadequate staff to locate each 
and every file or record that has been requested by Regional 
Offices across the country.
    Another significant issue, which can be identified at every 
Regional Office around the country, is the varying levels of 
experience of the Rating Veterans Service Representative. In 
any given case, you could take five individual RVSRs and give 
them the same file and come up with five different opinions on 
how the case should be rated.
    Although there have been improvements with the 
implementation of Rating Board Automation (RBA) 2000, the 
current electronic system utilized to rate compensation claims, 
the system is far from perfect. The overall ratings decision, 
including the service connection and actual percentage, is left 
up to the interpretation of the individual RVSR.
    The gap in varying decisions nationwide can also be 
attributed to the local policy at each individual Regional 
Office. While this has been the case for many years, the issue 
has come to a head due to the increased frequency at which this 
generation of veterans speak to each other and compare their 
individual situations.
    Mr. Chairman, unfortunately, there are--these are only a 
few of the issues that surround a paper-based system. And 
situations like the Clevelands' are not unique. Many working 
groups, Government Accountability Office reports, and 
commissions have made recommendations on this topic.
    Most recently, the Veterans' Disability Benefits Commission 
suggested that cycle times and accuracy could be improved by 
``establishing a simplified and expedited process for well-
documented claims, using best business practices and maximum 
feasible use of the information technology.''
    While the availability of well-trained, customer-service-
minded employees cannot be overvalued, the implementation and 
recommendations such as these can help to greatly reduce the 
complexity of the claims processing system and result in a 
timely--result in timely results.
    WWP looks forward to working with you and the VA to try to 
resolve these problems. Thank you again for the opportunity to 
testify today. And I will be happy to answer any questions you 
may have.
    [The prepared statement of Mr. Roberts appears on p. 47.]
    Mr. Hall. Mr. Roberts, thank you very much for your service 
to our country, the Marines, VA, and also now with the Wounded 
Warrior Project.
    And, I would start, I guess by asking Sergeant and Mrs. 
Cleveland what would you say were the biggest missteps in where 
the VA communicated with you?
    Mrs. Cleveland. That is just it--the lack of communication.
    Mr. Hall. Okay. It is just a simple answer.
    Mrs. Cleveland. Right. What happens is you just get a 
general form letter that says ``your file is incomplete'' or 
``medical records are necessary.''
    But then when you contact someone or you finally are able 
to get someone on the phone, they have no idea what it is that 
you are talking about. And it becomes submit or resubmit the 
entire package.
    Mr. Hall. So you were initiating most of the 
communications?
    Mrs. Cleveland. Exactly.
    Mr. Hall. Is this your file, by the way, on the table?
    Mrs. Cleveland. This is a part of it.
    Mr. Hall. It is----
    Mrs. Cleveland. This is a snapshot of it. And I was in the 
process. And I had it in one-inch binders is what it started 
out in. And it has grown quite a bit.
    Mr. Hall. The average we hear today is 183 days to process 
a claim. And that is hard enough to imagine. But in your case, 
it sounds like it went closer to 365 days.
    Mrs. Cleveland. Probably a little bit further than that, 
because we initially applied while he was still on active duty. 
July 2005 was when his application went in, the BDD.
    Mr. Hall. Well, somebody from this government ought to 
apologize to you. So let me be the first--if nobody else has, I 
apologize to you both on behalf of your government that you 
weren't taken care of and your needs were not attended to more 
quickly.
    Mrs. Cleveland. Thank you.
    Mr. Tai Cleveland. Thank you.
    Mr. Hall. I am sorry that happened. And we are going to try 
to make sure that it does not happen to future veterans 
anymore. I am going to try to reduce the time and reduce the 
number of repetitive requests, and stop making our veterans 
jump through hoops and prove that something is service related 
when it obviously is, and try to get people like you back 
integrated into something approaching normalcy and going about 
their lives in a much quicker way.
    My understanding was it took a full year for the VA to get 
you a check. And that was even after you underwent the BDD 
process.
    Mrs. Cleveland. Correct.
    Mr. Hall. What would have helped make this a better 
process, other than better communication? What would you list 
as the things that would have made it a better process for you?
    Mrs. Cleveland. If the process were fully automated, that 
would make a huge difference, because then you would not have 
to venture out on this paper chase.
    Mr. Hall. Right.
    Mrs. Cleveland. From my understanding, the file moves from 
one person to the next person in the rating process. And if one 
piece of paper ends up missing, the next person, it is 
something that they need, they don't--it is not as simple as 
going back and saying, excuse me, you just gave me this record. 
And--page 20 is missing. Can you locate it?
    Mr. Hall. Yeah.
    Mrs. Cleveland. It becomes the claimant's, the veteran's 
job to get that page 20 in there. Only they don't know it is 
page 20, so it becomes resubmit.
    Mr. Hall. Thank you very much.
    Mrs. Cleveland. The automation.
    Mr. Hall. Let me just ask Mr. Roberts, as a former Regional 
Office supervisor, could you describe for us how you would 
change this system to make it more effective and efficient for 
veterans?
    Mr. Roberts. Well in the Clevelands' case specifically, the 
VA historically, until the War on Terrorism started, they 
didn't--they were not used to taking active duty 
servicemembers, and taking claims while they were still on 
active duty, and establishing a claims folder.
    In their case, I would imagine that because he was on 
active duty, a claim folder was not established. Papers that 
were submitted, claims that were submitted, were not tracked in 
any way, shape, or form, and misplaced, lost. And that is why 
they were resubmitting over and over.
    The current claims processing system right now that--Mrs. 
Cleveland is absolutely right. It goes from one hand to 
another, from one team to another. And if the veteran has an 
appeal pending, then it could be in any team within the 
Regional Office at any given time.
    Definitely having the electronic file back and forth with 
DoD and VA would be the most beneficial system.
    Mr. Hall. Do you believe it is really necessary for six 
teams to handle one case?
    Mr. Roberts. No. This is--CPI was put into place several 
years ago. They used to have a team concept where files were 
rated. Everything was done within the same team. And the file 
stayed within that team.
    The way they have it set up now, everybody is doing part of 
the assembly line process. And they have their own specific 
part. And then it is passed on to the next person to do theirs. 
So it was a little bit easier years ago to do the claim, 
because you have RVSRs. You had decision review officers. You 
had veteran service representatives. You had all the components 
to work the claim right there on one team.
    Now responsibility gets passed along to whoever takes over 
after they get done with their part. And they pass it on to 
someone else. So it is hard to track.
    Mr. Hall. Thank you, sir. My time has run out. So I just 
want to ask very quickly--You mentioned that you can have more 
than one RVSR rating a case. Five different ones that come up 
with five different opinions----
    Mr. Roberts. Right.
    Mr. Hall [continuing]. Can VETSNET fix that problem?
    Mr. Roberts. VETSNET has come out. And they are working in 
it. Before I left the VA, it was just getting rolled out and 
being utilized. It doesn't fix it.
    And it is still the interpretation portion that the RVSR 
actually does on their own. They look at it. They make a 
judgment call based on the medical evidence. And based on their 
background, their experience, they make their decision. So it 
is still flawed in the current way it is rolled out.
    Mr. Hall. Thank you, sir. Now I will turn to our Ranking 
Member, Mr. Lamborn, for 5 minutes.
    Mr. Lamborn. I thank you, Mr. Chairman. Ms. Cleveland--Mrs. 
Cleveland, you mentioned that there are some unresolved issues. 
And the Chairman may have asked you briefly about that.
    Is there still anything as we sit here that needs to be 
resolved that we can help you with? Briefly; if not, we might 
have to talk separately or if you haven't already talked.
    Mrs. Cleveland. Separately.
    Mr. Lamborn. Okay.
    Mrs. Cleveland. Thank you.
    Mr. Lamborn. Okay, okay. Thank you.
    Mr. Roberts, do you believe that several of the problems 
that you laid out in your testimony to date could be solved 
with the new and up-to-date system, electronic system, that 
uses some form of artificial intelligence to adjudicate the 
claims?
    Mr. Roberts. Honestly, sir, I am just not that familiar 
with it. I wouldn't even want to get involved with that. And I 
will leave that up to the experts.
    Mr. Lamborn. Okay. Well, thank you for your candidness 
there.
    Why does the VA have a policy to place the responsibility 
to replace a lost file on the claimant?
    Mr. Roberts. Well, once the file is lost, they have no 
other option. And they are hoping that the claimants themselves 
have copies, like the Clevelands, in their possession. And they 
can resubmit and kind of rebuild the folder from the ground up 
again.
    You have to remember when the file is lost and they have to 
go through this process, they lose all service medical records, 
DD-214's, the initial claims, any medical evidence submitted 
from private physicians or medical facilities. Everything is 
gone. They have to start completely from scratch and rebuild 
the file from the ground up.
    Mr. Lamborn. Okay. And you said in your testimony that 
numerous pieces of vital evidence are often lost or misplaced 
and cannot be associated with the appropriate claim folder. 
Could you give us a little more specificity on how often you 
think this happens?
    Mr. Roberts. Well, I imagine--and just for an example, I 
used to use Houston, because I worked there. If the claim file 
is not where it is supposed to be, if it is not in the filing 
cabinet, or it is not at the person's desk that says it is 
actually located with, the mail is just put on search. And it 
is put in a bin in numerical order. And it sits there until 
somebody COVERS in a file to themselves and sees, you know, 
mail search pop up. And then they physically have to go get up 
and go get the mail and then associate it with the file. If 
they don't use the system, they never know the mail is there.
    I have seen files go all the way through the processing 
system, be adjudicated, be rated, be finalized, letter has gone 
out to the veteran, the file goes back, gets covered into the 
filing cabinet, and the little GS-4 file clerk goes, ``Oh, 
there is mail for it.'' And the process starts all over again. 
They have to go back and re-adjudicate and re-rate that claim 
based on the new evidence.
    The system they have now is the human error. If they don't 
use it, it doesn't do you any good.
    Mr. Lamborn. Now it sounds like some of the issues we are 
talking about right this minute, and in response to earlier 
questions from the Chairman, and based on the testimony from 
the witnesses, has to do not so much with artificial 
intelligence or how the claims are adjudicated, but how the 
records are stored, and kept, and processed, and transferred.
    So at a minimum, it sounds like we should be looking at 
digitizing some of these records to hopefully reduce the 
examples where things are lost and the time is lost trying to 
retrieve them, if that can be done. Or multiple people can look 
at them at the same time, if we have these six teams, more than 
one of which might be looking at it at the same time.
    Do you think that is a step that the VA, at a minimum, 
should take?
    Mr. Roberts. I think that is exactly what they should be 
doing. I have seen--this is a small example of files I have 
seen. I have seen two or three boxes just for one file, one 
veteran, in large boxes. And I have seen boxes get lost that 
belong with other boxes.
    So I got--I have seen files from veterans filed--half of 
the file in one side of the building and the other half of the 
file on another side of the building. And it takes--I have seen 
up to a month for them to actually connect the two of them 
together.
    So, yeah, I have actually seen people on the appeals team 
working a claim with half a file. And people in 
predetermination working on half a file. And I am not sure how 
they did it. But I have seen it.
    Mr. Lamborn. I thank you for your testimony. Mr. Chairman, 
I yield back.
    Mr. Hall. Thank you, Mr. Lamborn. It would be funny were it 
not so serious. The Chair will now recognize Congressman 
Bilirakis for 5 minutes of questioning.
    Mr. Bilirakis. Thank you, Mr. Chairman. My question has 
already been answered.
    But I appreciate you holding this hearing. And we need to 
solve this once and for all, because I know it has been going 
on a long time. And the claims--the process is too long.
    Thank you very much for testifying today. And thank you for 
your service.
    Mr. Hall. I would add my thanks to all of you, and just say 
that, Mr. Roberts, the help that the Wounded Warrior Project 
provided and you provided is very welcome I'm sure to the 
Clevelands but also to all of us.
    And, just knowing that this is not an isolated incident, I 
hope that we can set up a system using as much digitizing, 
electronic storage, and electronic motion, and shared files, as 
the Ranking Member was saying, so that we can avoid this; what 
looks like it is well over a foot high. If you piled those on 
top of each other, a foot high, for what you are saying is a 
relatively small case in terms of the amount of information.
    But at any rate, thank you for your testimony.
    And we have votes that are under way now. So we will recess 
the hearing for as long as it takes for us to go across the 
street and vote. When we come back, we will hear from our 
second panel.
    This Committee stands at recess until then.
    [Recess.]
    Mr. Hall. The Subcommittee will come to order. Thank you 
for your patience. We now have joining us at the witness table 
panel two.
    Dr. Tom Mitchell, Chairman of the Machine Learning 
Department, School of Computer Science at Carnegie Mellon 
University; Dr. Randolph Miller, Chairman of the Department of 
Biomedical Informatics at Vanderbilt University School of 
Medicine; Dr. Marjie Shahani, Senior Vice President of 
Operations at QTC Management, Inc.; Mr. Ned Hunter, Chief 
Executive Officer from the Stratizon or is it Stratizon?
    Mr. Hunter. Stratizon.
    Mr. Hall. I am thinking of that other company that ends 
with ``izon'' Corporation, to describe a pilot study in 
Virginia. Mr. John F. McGarry, Senior Vice President of 
Benefits and Chief Risk Officer at Unum; and Mr. Gary 
Christopherson, the former Veterans Health Administration Chief 
Information Officer, former Senior Advisor to the Under 
Secretary for Health, and former Principal Deputy Assistant 
Secretary for Health Affairs. A distinguished group indeed. 
Ladies and gentlemen, welcome to this Subcommittee. Your full 
written statements have been entered into the record. And so 
feel free to cut corners if you wish so that we will have time 
for questions.
    We are expecting to have Ranking Member Lamborn back here 
any time. But since the next round of votes is scheduled in 
about 40 minutes, we are going to try to move this along so we 
can hear from you and not interrupt the panel to have to go 
vote.
    Mr. Mitchell, you are now recognized for 5 minutes.

STATEMENTS OF TOM M. MITCHELL, PH.D., E. FREDKIN PROFESSOR AND 
CHAIR, MACHINE LEARNING DEPARTMENT, SCHOOL OF COMPUTER SCIENCE, 
CARNEGIE MELLON UNIVERSITY, PITTSBURGH, PA; RANDOLPH A. MILLER, 
M.D., DONALD A.B. AND MARY M. LINDBERG UNIVERSITY PROFESSOR OF 
   BIOMEDICAL INFORMATICS, MEDICINE, AND NURSING, VANDERBILT 
 UNIVERSITY SCHOOL OF MEDICINE, NASHVILLE, TN; MARJIE SHAHANI, 
M.D., SENIOR VICE PRESIDENT, OPERATIONS, QTC MANAGEMENT, INC., 
 DIAMOND BAR, CA; NED M. HUNTER, PRESIDENT AND CHIEF EXECUTIVE 
  OFFICER, STRATIZON CORPORATION, ATLANTA, GA (VA STATE PILOT 
  STUDY); JOHN F. MCGARRY, SENIOR VICE PRESIDENT OF BENEFITS, 
      CHIEF RISK OFFICER, UNUM, PORTLAND, ME; AND GARY A. 
 CHRISTOPHERSON, UNIVERSITY PARK, MD (FORMER SENIOR ADVISOR TO 
THE UNDER SECRETARY FOR HEALTH, AND CHIEF INFORMATION OFFICER, 
  VETERANS HEALTH ADMINISTRATION, U.S. DEPARTMENT OF VETERANS 
 AFFAIRS, AND FORMER PRINCIPAL DEPUTY ASSISTANT SECRETARY FOR 
          HEALTH AFFAIRS, U.S. DEPARTMENT OF DEFENSE)

              STATEMENT OF TOM M. MITCHELL, PH.D.

    Dr. Mitchell. Thank you Chairman Hall and distinguished 
Members of the Committee.
    It is an honor for me to be asked to testify here today, 
and to try to help you help the members of our armed services 
who have served.
    Clearly, we face a significant problem and backlog in the 
processing of benefits claims by the VA. In my opinion, we have 
the technology needed to address and to eliminate this problem. 
Think for a moment of the forms filling problem that we are all 
familiar with, filling out forms for income taxes.
    If we can develop computer software like TurboTax, which 
helps us fill out very complex multiple page forms, guides us 
through the steps to determine what kind of information to put 
in which kind of field, and then can instantly apply very 
complex tax regulation codes to calculate to the penny the 
amount of income tax that we owe, then I don't see why we can't 
develop software that performs an analogous function for the 
people who have to fill out forms for VA benefits and the 
people who have to apply the complex regulations to those.
    To take a second example that is even more similar to the 
problem faced by the VA, consider the current practices for 
processing benefits claims in the medical insurance industry.
    At Highmark Inc., which is a major provider of health 
insurance in my home State of Pennsylvania, I am told that 90 
percent, nine zero percent, of the claims that come in from 
physician offices and from hospitals are automatically 
processed without any human intervention.
    How do they do this? They do it by using electronic forms 
instead of paper. They do it by coding the treatments that the 
patients have received using industry standard (International 
Statistical Classification of Diseases and Related Health 
Problems) ICD-9 codes. They do it by developing rule-based 
software that captures the rules and regulations by which the 
correct payment is calculated from the details of the treatment 
received by the patient.
    And after the decision is made automatically by the 
software, the payment is issued automatically. So that process 
happens in 90 percent of the cases automatically. And the other 
cases require human intervention.
    Can the VA do the same? While the type of benefits claims 
processed by the VA may be somewhat different from those in the 
medical insurance industry, it seems to me the problems are 
similar enough that we ought to expect that the VA can also get 
a great benefit out of this kind of automation.
    In my opinion, it is useful to consider a three-stage 
introduction of computer technology for claims processing in 
the VA. First, we can shift from pencil and paper claims to 
online claims. This alone would improve the accuracy, 
efficiency, and as we heard in the previous panel, the ability 
to hold onto and not lose claims.
    Second, introducing computer software to help interpret 
these online claims to apply the regulations about which 
benefits are due would be a second step. We have well 
understood technologies for encoding complex regulations in 
software such as rule-based systems.
    And for steps that require some human subjective judgment 
along the way, we also have technologies such as case-based 
reasoning, which allow the computer to pull up the two or three 
most similar previous claims in the system for inspection by 
the human as they are applying their judgment to this new case.
    As the third step, once these claims are online and the 
processing is automated, the resulting database of claims can 
itself serve as a resource for data mining. Data mining methods 
can be applied to the claims data.
    For example, data mining can be used to predict and flag 
new claims that are outliers that might require some 
specialized expertise to evaluate them, or to identify 
soldiers, veterans, who should be taking advantage of services 
that they appear not to and alerting them.
    So to summarize, in applications from insurance claims 
processing to tax filing to customer help centers, there is a 
growing and widespread use of computer-based tools for 
capturing data in forms and for applying automatic rule-based 
inference to those.
    Much of this technology comes out of research previously 
sponsored by Federal agencies such as the National Science 
Foundation and Defense Advanced Research Projects Agency. But 
the core technology is by now very well understood. This is not 
bleeding-edge technology.
    The VA should take advantage of this. And I recommend three 
steps that can be carried on in parallel to get started.
    One, conduct a detailed 3-month study of the workflow 
process in the benefits office to determine the different steps 
and to identify for each of those steps whether it can be 
automated. If not, whether some computer support such as case-
based reasoning can be used to help in the human judgment.
    Second, begin immediately to move all of the claims online. 
Even without any additional processing, just having them online 
will be a benefit.
    And third, consult with large insurance companies and 
others who process benefits claims more automatically to 
understand what are the current best practices and to begin a 
process of adopting those where appropriate.
    Thank you, Mr. Chairman, for your attention and for the 
opportunity to address the Committee.
    [The prepared statement of Dr. Mitchell appears on p. 49.]
    Mr. Hall. Thank you, Dr. Mitchell.
    Dr. Miller, you are now recognized for your opening 
statement.

             STATEMENT OF RANDOLPH A. MILLER, M.D.

    Dr. Miller. Thank you, Mr. Chairman, for the opportunity to 
address the Subcommittee this afternoon.
    My comments describe the applicability of biomedical 
informatics to the processes determining veterans' eligibility 
for disability compensation.
    Clinical informatics involves application of computer-
assisted technology for information management and 
decisionmaking during healthcare delivery.
    If I could have my slide over here. Don't worry, I am not 
going to read the whole slide. So the problem we have at hand 
is first the criteria in CFR 38 part 4 are vague and ambiguous. 
For example, in section 56, part C, muscle disability is 
defined as ``loss of power, weakness, lower threshold of 
fatigue, and fatigue pain.''
    While I can't do as many push ups as I did when I was 20, I 
can't run the mile like I used to, and they talk on Sundays in 
the NFL broadcast about the athletes working through the pain 
of fatigue, I do not consider myself or pro athletes disabled. 
And so the criteria are very ambiguous.
    So the first thing is for Congress to redefine what they 
really mean in a way that is actionable. Otherwise, computers 
won't be able to help.
    Another key principle of informatics is that you need to 
identify the most proper, correct, definitive source of 
information, collect information from that source, once and 
only once, and record it once in a place where everybody else 
can access it without overriding it with incorrect information.
    So in addition to the veteran himself or herself, there are 
three places of major activity relative to disability 
determination. During active duty, when somebody is injured or 
wounded, they should collect disability information right 
there--beginning at the time that the service man or woman 
receives care, and collect it in a way that is relevant to 
disability claims, so that doesn't have to be replicated later.
    After discharge, the veterans are seen within the VA 
healthcare system, and they should collect disability 
information there. The Compensation and Pension Record 
Interchange (CAPRI) system is the beginning of a good way to do 
that. But it is only used on about 25 percent of disability 
examinations now.
    And then finally, as we have already seen in the previous 
panel, there is more than ample opportunity to automate the 
paper records system for VBA.
    And in my written statement, I presented three different 
layers, starting with simple collecting of information to more 
complicated things like AI applications that can be used to 
progressively refine the system.
    And I would also like to point out, as Dr. Mitchell stated, 
that once all of this information is automated, not just in 
scanned records but in actionable form, then you can collect 
information about which claims are more difficult to process or 
take longer time, which Regional Offices are efficient and not, 
which veterans need more attention because they haven't been 
processed yet, and so on.
    When everything is electronic, you can do quality 
improvement much more effectively than you can with paper.
    As I have stated, and the Chairman pointed out in his 
opening comments, artificial intelligence and expert systems 
cannot replace human intelligence and human compassion in 
judging whether veterans qualify for disability benefits. But 
they can speed up the process and help the VBA make it more 
uniform and more accurate.
    It is very important to realize that you can cause problems 
by automating things as well as curing problems. So, for 
example, if in the process of implementing improvements the VBA 
raters had a half electronic system and half paper system, they 
would never know whether information was in the paper side or 
the electronic side. And they would have to go to both all of 
the time.
    So this needs to be done in a thoughtful way, where people 
are helped at each step and the situation is not made more 
chaotic or confusing. And it needs to be done in a 
nondisruptive manner.
    The way the VA has implemented the VistA system is 
exemplary nationally in informatics. And that would be a good 
basis on which to model future changes. Thank you.
    [The prepared statement of Dr. Miller appears on p. 53.]
    Mr. Hall. Thank you, Dr. Miller.
    Dr. Shahani, you are now recognized for 5 minutes.

               STATEMENT OF MARJIE SHAHANI, M.D.

    Dr. Shahani. Mr. Chairman, Members of the Subcommittee, 
thank you for the opportunity to testify before you today on 
the important topic of processing veterans' claims.
    QTC is a nationwide provider of medical examinations and 
record review services to the medical and disability 
communities. We actually support Federal, State, local 
government agencies; property and casualty insurance carriers; 
third-party administrators; employers and the claimants they 
serve.
    We have been a provider of compensation and pension medical 
examinations services to the Veterans Benefits Administration 
since 1998.
    QTC provides the detailed medical examination for veterans 
and then submits the exam report to the VA's claims 
adjudicators or rating specialists who then, along with the 
veterans C-file or claims file, rates the veteran's disability 
claims.
    To ensure a quality, timely, customer-focused, and cost-
effective process and medical report, QTC pioneered the use of 
software and technology. In every step of our process, we have 
created software to facilitate and improve our own efficiency.
    Over our 9 years of experience working with the VA, we have 
come to understand the unique and complex challenges of the VA 
disability process. It is like no other disability program with 
which we work.
    In an attempt to provide value-added services to VBA and 
for veterans, QTC applied its knowledge and experience 
specifically to simplify and streamline the information 
gathering process for VA's rating specialists.
    QTC actually developed what we call an Evidence Organizer 
prototype. It is an automated tool designed to assist VA's 
rating specialists significantly reduce the time to determine a 
rating decision.
    The Evidence Organizer has great potential in helping 
rating specialists search and find relevant medical information 
critical to make that final rating decision.
    How does it work? Basically it converts the cumbersome 
paper-based c-file to create an electronic record or e-file. I 
guess that is what everybody is saying. First we have to 
convert the paper into something electronic.
    This document management process begins with a technician 
scanning in the entire c-file through the use of optical 
character recognition. The software transforms each record into 
a text searchable digital record.
    At the heart of this process is QTC's core knowledge 
database, which is built upon our extensive disability 
examination experience supporting the VA's Compensation and 
Pension exams.
    The knowledge database identifies, highlights, and 
electronically indexes all keywords. For example, claimed 
conditions like diabetes, asthma, arthritis, as well as any 
potential claimable conditions throughout each medical record.
    Once the e-file has been established, each record is 
reviewed, validating the software's indexing, highlighting the 
records, and now actually linking the referenced medical 
records and evidence in the c-file to VA's rating requirements 
or rating codes.
    Once all medical records have been reviewed and linked, the 
e-file is now ready for VA's rating specialist. Right now as we 
understand it, the c-file is organized or filed according to 
the date reports or documents are received.
    In addition, most rating specialists process a veteran's 
case addressing and rating one claim condition at a time. Thus, 
in addressing a veterans' case with four claim conditions, the 
current average, the rater reviews the entire paper claims file 
repeatedly, making notes, putting sticky notes, clipping files 
together to organize the medical evidence.
    The Evidence Organizer will not only organize the medical 
evidence by claim conditions, but also link the available 
evidence to the actual rating requirements, allowing the rating 
specialist to still make that final determination and write the 
rating decision.
    Upon consultation with former VA rating specialists, we 
estimate that turning this manual paper process into an 
electronic process will actually improve productivity by 37 
percent per decision. By applying technologies such as the 
Evidence Organizer to this paper process, VBA could greatly 
reduce routine and repetitive administrative tasks for rating 
specialists, improve their efficiency, and ensure quality and 
accuracy of each review.
    Thank you again for the opportunity to testify this 
afternoon.
    [The prepared statement of Dr. Shahani appears on p. 58.]
    Mr. Hall. Thank you, Dr. Shahani.
    Mr. Hunter, you are now recognized for 5 minutes.

                   STATEMENT OF NED M. HUNTER

    Mr. Hunter. Chairman Hall and distinguished Members of the 
Subcommittee, thank you for the opportunity to appear before 
you today.
    Stratizon Corporation is a veteran-owned Software-as-a-
Service company, which has utilized the concepts of artificial 
intelligence to successfully design a software platform and 
application solely focused on improving the VA claims 
processing system.
    We have gained valuable insight into the underlying success 
of using AI to solve the VA claims processing system. First, 
the technology available is in the marketplace. It is 
adaptable, flexible, scalable, proven, and cost effective. 
Technology is not to be resisted but embraced.
    Second, success will be highly dependent upon the 
perspective in which AI solutions are constructed. A true 
veteran-centric solution of the future must be constructed 
through the eyes and the situation of the veteran to satisfy 
the requirements of the State and Federal policies and VA 
systems and not constructed through the eyes of the multiple 
government entities to independently present the bureaucracy to 
the veteran.
    Stratizon applied this perspective in successfully piloting 
for the United States Navy, three unique web-based intelligent 
solutions that demonstrated how the quality of life for sailors 
could be significantly improved by replacing confusing, 
complicated, paper intensive, and manually driven enterprise 
processes with web-based, easy-to-use, fully automated, and 
complete self-service solutions, or what we define as 
``intelligent user interfaces'' or ``IUIs.'' And our tool does 
this without the use of any programs or hard coding.
    IUIs can also be designed for numerous veteran events such 
as transitions from active to veteran status or applications 
and appeals for VA compensation and health benefits.
    The Commonwealth of Virginia's Department of Veteran 
Services, working with the Joint Leadership Council of Virginia 
representing 32 veteran service organizations, is implementing 
such a solution called TurboVetTM.
    Building on a successful pilot in 2007, the Governor of 
Virginia has included funds in his fiscal 2009 budget that 
begins on July 1st, 2008, for full production.
    TurboVetTM will provide Virginia veterans, or an 
authorized representative, or survivor the ability to log 
online at Virginia.gov, via a personal computer or device such 
as this Apple IPhone, and select an event that they need 
assistance with.
    Initially a series of statements and questions regarding 
their status or particular event will be presented. Their 
personal data currently on file with the State will be 
retrieved so they may confirm or validate that data, thus 
improving data integrity and eliminating redundant data entry.
    The system will use embedded decision logic to react 
intelligently to their input to continually refresh and display 
only the necessary event questions, thus eliminating the 
frustration of redundant and unnecessary questions.
    A list will be displayed of all State and Federal benefits 
the veteran has earned with all corresponding documents 
spanning multiple agencies required for the veteran to submit, 
thus providing a peace of mind to the veteran their solution is 
holistic.
    Each document will then be progressively, simultaneously, 
and perfectly auto-populated with the proper data, thus 
eliminating data transcription errors and numerous processing 
delays.
    Finally, the veteran will have the option to save and print 
each document locally and, at their discretion, electronically 
submit their data securely to all participating authorities and 
systems to be processed and tracked fully and completely.
    Virginia's success in using an AI platform is dependent 
upon the continued support and cooperation of all parties, both 
political and technical. Decisionmakers need to remain 
committed to this paradigm shift to the future and must always 
provide their best institutional knowledge available to ensure 
the TurboVetTM IUI not only becomes that benchmark 
of service but also remains that benchmark.
    We need technical cooperation between State agencies to 
take advantage of TurboVetTM's ability to seamlessly 
exchange data with disparate IT systems. We need cooperation 
and support at the Federal level.
    Federal supervisors in Roanoke have projected that a 
minimum of 100 days of processing time will be eliminated when 
the TurboVetTM system is implemented at only the 
State level.
    Stratizon foresees few problems in exchanging data between 
TurboVetTM and VA systems such as VistA and VETSNET. 
We fervently believe there could be significant process cycle 
time improvement and extraordinary cost savings at the State 
and Federal level if veteran's data at the State level could 
first be pre-verified against recognized authoritative national 
VA databases and then seamlessly exchanged upon claims 
submission and during the claims management process. Virginia's 
goal is to fulfill the vision of House Resolution 3047 and have 
a claim prepared properly with attached medical evidence and 
documentation for electronic submission to Federal adjudicators 
for rating, and have those claims calculated fairly, 
consistently, and automatically.
    In summary, using a properly designed AI system would 
dramatically improve the VA claims processing systems by 
improving the access to customer solution and service for 
veterans and their family members, reducing the costs to the 
State in staff administration, training, and paperwork, and 
improving the accuracy, throughput, and expediency of claim 
submissions by the State for VA adjudication.
    On behalf of the Stratizon Corporation, I would like to 
thank the Chairman and all Subcommittee Members for this 
opportunity to be here today.
    [The prepared statement of Mr. Hunter appears on p. 61.]
    Mr. Hall. Thank you, Mr. Hunter.
    Mr. McGarry, you now are recognized for 5 minutes.

                  STATEMENT OF JOHN F. McGARRY

    Mr. McGarry. Mr. Chairman, Members of the Subcommittee, I'd 
like to thank you for the opportunity to testify before you 
today.
    My name is Jack McGarry. I am the Senior Vice President of 
Benefits and Chief Risk Officer at Unum.
    I have submitted written testimony, which has been made 
available to you. But will briefly present an overview.
    I am here today to discuss how our technology facilitates 
claim management decisions at Unum. We process approximately 
400,000 disability claims per year and pay about $4 billion in 
benefits directly to our insureds and their families.
    Most of Unum's claims are governed by the Employee 
Retirement Income Security Act (ERISA), the Federal law which 
generally requires insurance companies to make disability claim 
decisions within 45 days.
    Unum's experience shows that it is possible to manage high 
volumes of claims in a timely and accurate manner while 
achieving high levels of customer satisfaction.
    Technology is an important component to the solution of 
managing volumes, timeframes, and customer service. However, 
the decision about a person's ability to work is also informed 
by in-depth analysis of pertinent documents and discussions 
with claimants, their employers, and their physicians in order 
to assess their ability and motivation to work.
    In the end, the disability determination is a judgment call 
that needs to be made by a person.
    In order to assure that the right people are reviewing the 
right claims at the right time, a combination of Unum's 
technology and people is necessary.
    For example, a routine claim may be automatically sent by 
the system to one person, while a complex claim with multiple 
diagnoses may go to another based on a combination of systems 
and management decisionmaking. As robust as our systems are, a 
person does look at every claim we pay.
    Our technology does the following. It manages documents, 
facilitates workflow, ensures a complete administrative record, 
and monitors and measures quality and service results.
    First, our system manages documents. Our files can grow to 
hundreds if not thousands of pages. With our image-based system 
all files are paperless and multiple people can access the 
claim same--same claim at the same time. Documents are 
organized and stored in an efficient manner.
    Second, our system facilitates workflow. All new documents 
and other information are electronically scanned into our 
system upon receipt. Our technology facilitates parallel claims 
processing and ensures claims issues are promptly addressed.
    The act of scanning the documents as they are received 
creates an online activity for the claim payer to review. In 
our system, every action a person completes creates another 
action or follow-up activity.
    The system can also trigger an action for someone to review 
claims and/or contact customers at key times during the claims 
management process.
    Third, our system ensures a complete administrative record. 
An administrative record is important for ERISA purposes as 
well as sound claim management.
    When a claim changes hands between claim payers, all of the 
management activities associated with that claim, including 
future activities, stay with the claim and are automatically 
assigned to the new claim payer.
    The technology keeps the file together in one place and 
minimizes any disruption in service due to a personnel changes.
    Fourth, our system monitors and measures quality and 
service results. Management and our quality assurance process 
require the ability to review files real time, at the same time 
that the claim payers are working on the files. The system 
automatically tracks and reports on service times and outcomes.
    At the initial level, for the shorter terms claims, our 
intake department reviews each new claim and assigns an ICD-9 
diagnosis code. Our technology then separates the levels of 
disability into those which have shorter durations and those 
which may be longer term based on the assigned diagnosis code.
    Simpler claims are triaged directly to a claim payer. For 
the most complex claims, our technology triages the claims to a 
manager who decides which claim payer to assign the claim based 
on the experience of the individual.
    After the initial assignment, our technology initiates 
reports based on key measures, including diagnosis, generally 
accepted medical condition guidelines, and our own Unum 
database information. These reports can identify claims that 
need additional work or follow up, and help each claim payer to 
determine what steps to take next.
    Disabilities present a complex management challenge, 
because they are logistically difficult, judgment based, and 
can be emotionally charged. Technology can help facilitate 
judgment-based decisionmaking, but we don't see it as ever 
being able to replace people in the claim management process.
    I would like to end by extending an invitation to all of 
you and for the VA staff to visit Unum and would welcome the 
opportunity to continue to be a resource for sharing best 
practices between the public and private sectors as you 
continue to evaluate the disability adjudication/case 
management process.
    Thank you for the opportunity to testify before the 
Subcommittee.
    [The prepared statement of Mr. McGarry appears on p. 64.]
    Mr. Hall. Thank you very much, Mr. McGarry. You and Dr. 
Miller have helped set standards for efficiency by finishing in 
under 5 minutes. Not that we will hold anybody else to that.
    Mr. Christopherson, you are next, and are recognized for 5 
minutes please.

              STATEMENT OF GARY A. CHRISTOPHERSON

    Mr. Christopherson. Chairman Hall, Mr. Lamborn, Members of 
the Subcommittee, let me applaud you for holding these very 
important hearings and for your opening remarks.
    Today I am going to speak to the enabling role of 
artificial intelligence, to the true obligation of duty to 
assist, and to the honor bestowed on those who deliver on time 
and on target.
    When I served as Principal Deputy Assistant Secretary of 
Defense for Health Affairs, I saw our servicemembers sacrifice 
and our Nation incurred debt.
    I saw our veterans' plight when serving as VHA Chief 
Information Officer and Senior Advisor to the Under Secretary. 
And I had the great privilege of getting to know 
servicemembers, veterans, and their support organizations as 
people providing a great service to our Nation.
    All this taught me that everything VA does should be 
centered around the veteran. It is not today. If we believe 
that veterans are hurting, and that we have the duty to assist, 
and that we should be on time and on target, we need a new 
claims system. And we need it now.
    When I was advising VBA in thinking about a new system 
several years ago, I learned it takes 6 months to a year or 
more to complete about 8 hours of actual work. Unacceptable.
    When a veteran is hurting and needs healthcare, the VA 
health system assists the veteran and provides care quickly. 
When a veteran is hurting and needs financial benefits, the VA 
benefits system does little to assist, forces the veteran to 
navigate a large bureaucracy and massive paperwork, and 
provides financial benefits only after months or years. Sadly 
this all happened to Gunnery Sergeant Cleveland.
    So what should happen? First place, VA staff should be 
coming out and welcoming the veteran, not the way it is done 
today. They should actively assist the veteran to get 
everything processed quickly and correctly. Longer term, they 
should assist as case managers.
    Further, we need the continuing and valuable support and 
assistance of the veterans' service organizations.
    In my opinion, changing the process means giving a veteran 
a temporary financial benefit at least as soon as the veteran 
files a claim with basic supporting evidence.
    For the permanent decision, real time would mean the VA 
could receive the claim with supporting evidence and make the 
decision on the same day or at least within a couple of weeks. 
Further, let us start paying the veteran within 30 days.
    In my colleagues' testimony, we heard that technology 
exists today to greatly improve the speed and accuracy of 
benefit decisions.
    For those who argue claims processing is a much more 
complicated and difficult process, I counter that it is not. 
Healthcare, much more complicated and difficult, is figuring 
out how to provide care in real time without technology and 
even better with technology.
    When I rescued the VistA health information system and 
moved it to a brighter future, we also made that information 
available to VBA electronically and in real time.
    Now artificial and human intelligence together can help. VA 
healthcare providers have the decision support to care well for 
a person in real time. For claims processing, we do not have to 
wait for the technology. We can start reducing the misery today 
and even better when the technology arrives. However, getting 
to a new, veteran-centric, effective claims processing system 
with the necessary enabling technology will only happen if VA 
leadership is fully committed to achieving that vision.
    Further, VA leadership will need effective management and 
staff to make all this happen.
    Yes, this is all affordable and doable. First, it could be 
well built into the $150 billion economic stimulus package 
moving at this very moment through the Congress.
    Secondly, we have to understand that we handle the budget 
when we send our servicemembers to war. We should do no less 
when they come home and need our help. This is a part of real 
cost of preventing or conducting war.
    Today, there is a failure to understand and appreciate the 
veteran's plight. Feel what it is like for a veteran to live in 
uncertainty and without support for months, or a year, or more. 
If we did that for healthcare, that would be totally 
unacceptable.
    Bottom line, change the assumptions. Change the process. 
Use the best technology. Change the attitude. Care for the 
veteran. On time and on target is what we expected of our 
veterans and what we should expect of VA. The duty to assist is 
an obligation that VA with regards to benefits has yet to 
honorably discharge.
    Thank you, Mr. Chairman.
    [The prepared statement of Mr. Christopherson appears on
p. 75.]
    Mr. Hall. Thank you, Mr. Christopherson.
    Eloquent, powerful testimony all. The question I wanted to 
ask, first of all, Mr. Christopherson, you talked about 
starting paying veterans once they filed a complete claim 
within 30 days. First you said immediately. And then you said 
at least within 30 days. This is something that many of us have 
been advocating for.
    Do you have a figure in mind or a percentage disability 
rating in mind that would be your best guess average or, you 
know, baseline to start while the process goes forward?
    Mr. Christopherson. Not really. I think what you have to 
look at the situation of the person's need, the veteran's need 
at that time.
    Secondly, there is obviously a political process you have 
to go through with budgetary decisions. And what I may ask for 
and what I think should happen, my sense of right now is if a 
veteran has a disability, whatever degree is appropriate at 
that time, that we have some degree of confidence in, grant it. 
And start paying it within 30 days in terms of that.
    Second part is, and by the way, that starts to shift the 
burden onto the VA rather than onto the veteran. Right now we 
have a backward, upside down system. Where we sort of say if 
you can figure out how to navigate the system, maybe we will 
let you get benefits. And Lord knows how long it is going to 
take.
    If we start the reverse and say we are going to start 
paying some benefits, and we will make some mistakes, but by 
the way, they served. We didn't essentially ask them a lot of 
questions at that time. They didn't demand a lot of answers at 
that time. We should be doing the same here.
    So I think essentially what you really have now is give as 
much as you can with a certain reasonable amount of risk. If we 
do it for a temporary basis and for a relatively short period 
of time, the government is not at great risk in doing that.
    If you couple that with all the things we have talked about 
here at this table about moving the whole time process down, 
the risk to the taxpayer goes down very substantially as well.
    Mr. Hall. Thank you. Mr. Mitchell, your work in artificial 
intelligence covers a broad area from computer learning to 
advanced robotics.
    I was wondering what level of technology are we talking 
about for transforming the VA claims processing system?
    Mr. Mitchell. Well I think the example of Highmark is a 
good one. They automate the claims processing. And, in fact, 
TurboTax is another good example. Both of these are systems 
that essentially are very well understood. These are not 
bleeding-edge technology. They are based on very well 
understood techniques that come out of artificial intelligence.
    But essentially they are ways of encoding in software a 
large collection of rules like the one that you mentioned in 
your own opening remarks that say ``if, the form has this kind 
of data, then this is an appropriate kind of disability 
rating.''
    And so that technology for rule-based processing is very 
well understood. It is something we could do today and is 
widely done today.
    Mr. Hall. How long would you guess it would take, Mr. 
Mitchell, for such a system to be created?
    Mr. Mitchell. I believe if--so I looked in preparation for 
this meeting at some of the rules that are used for assigning 
disability benefits based on these conditions. To take a 
standard rule-based engine and to input those kind of rules is 
months. It is well under a year.
    Now I can't estimate the additional sort of organizational 
and bureaucratic adjustments that would be needed--that would 
have to be done to go along with that. But from a purely 
technical standpoint, we are talking about months.
    Mr. Hall. First to you, and then to anybody else on the 
panel, how important would it be that we get a digital handoff 
from the DoD to the VA? I heard when I was in Landstuhl, 
Germany, in October on my way back from Iraq. The commander who 
is in charge of the hospital in Landstuhl says that they are 
bringing back the servicemembers who are injured with an 
electronic record, which is like an onion. They keep adding 
another layer to the onion at each place to what they added in 
theater.
    And then they added in, the treatment they are getting 
while they are being flown, and then when they get to 
Landstuhl, they add more records about the medications, or the 
therapies, or the treatments, or surgeries, whatever is 
happening to that veteran.
    When they return to Walter Reed or Bethesda, then another 
layer is added to the onion. They told me that in December, 
last month, they were going to be able to start handing this 
off to the VA. Well, I am not sure if--I haven't gotten a clear 
answer as to whether this is actually happening yet. But 
assuming that happens, how important is that to being able to 
start this process?
    Mr. Mitchell. Yes. You know, I would leave that to people 
who know more about the detailed decision--the detailed policy 
for assigning benefits. But it is clearly the case that these 
rule-based systems can apply only to data that has already been 
captured online.
    And so if that part of the electronic record is relevant, 
then it would have to be online, either by being passed off or 
by being transcribed from paper in some other way.
    Mr. Hall. Dr. Miller.
    Dr. Miller. The CAPRI system that the VA has developed for 
examiners to record the disability exams within VHA to pass 
along to VBA, I believe it was already in pilot that you are 
referring to. So the problem is the DoD records are in 
different format computationally than the VA records are. And 
that is one of the logjams in the disability determination.
    But for the BDD process that Dr. Christopherson referred to 
it, if in active service they use CAPRI forms to do the quick 
and dirty disability determination, that is an existing system 
the VA developed. And they could probably use that as the basis 
fairly quickly for the initial short-term disability ranking 
and payment while more electronic work is done.
    Mr. Hall. And----
    Mr. Christopherson. Mr. Chairman, if I may.
    Mr. Hall. Yes?
    Mr. Christopherson. Let me fill in. It started when I was 
at DoD and then continued when I was at VA, which was the idea 
of doing exactly what you are describing. Which is to make the 
information that DoD generates electronically available to VA 
both for healthcare and for benefits determination there.
    Much of that information is now available. If it is 
electronic, it is available to VA both sides of the equation 
there. What you have to look at, what will slow things down is 
for older veterans who didn't have much care electronically in 
DoD. It has to go through the paper route.
    The later era, you have a mixed bag of that. You have to 
sort of deal with the mixed bag of that. But again, digitalize 
that and then essentially move it across.
    The next generation coming through should be heavily 
digitized. And the data should be standardized, which means you 
really can feed it into the rules engines that these folks are 
talking about here.
    Mr. Hall. So the most time-consuming task that we face is 
entering all of the old data that is in boxes and files, like 
the ones we saw earlier, into the system. And then starting 
from whatever point the system is online, hopefully it will be 
expedited and more or less instantaneous.
    Mr. Christopherson. I would suggest, Mr. Chairman, that is 
not a staff issue. That is more likely going to be a contract 
issue. You can make that happen as far as you are willing to 
spend money to make it happen.
    Mr. Hall. It always comes down to money, doesn't it? Dr. 
Miller, I have one more question for you. I have often heard 
that doctors use a technique called differential diagnosis 
where they have a hypothesis about a patient's illness and then 
ask questions to rule out conditions until they come up with a 
diagnosis.
    Can a computer using a rule-based expert system as you 
described, assist with assigning disability ratings that cover 
the VA's 700 codes and its zero to 100 percent range of 
severity that often includes multiple conditions? How long 
would it take a computer to do that?
    Dr. Miller. I worked on diagnostic systems of the type you 
are referring to for a quarter century. There are probably 
seven or eight techniques in addition to rule based that can be 
used to do what you have asked.
    Essentially the idea has already been stated. But you would 
use electronic means to identify findings in the veterans 
records or an active service person's records. And that could 
cue the practitioner taking care of them that this patient is 
potentially eligible for disability and hone down into the 
specific categories of the 700 that the veteran might be 
eligible for.
    In the end, it still should be a decision by a human. But 
reminding people when they might not be thinking about 
disability in the heat of battle or whatever that is an 
important component of the care is something that such tools 
would be able to do.
    Mr. Hall. Thank you very much. Dr. Shahani, we have often 
heard that claims have become more complex with over eight 
conditions per claim instead of just one or two.
    Could a system such as QTC's rate all of those conditions 
given that the claim is already in a ``ready to rate'' format 
such as the one described by Mr. Hunter? How long would that 
take?
    Dr. Shahani. Just to paraphrase your question again, are 
you asking then for the time it would take to code all the 700 
codes, the multiple conditions?
    Mr. Hall. Yes.
    Dr. Shahani. Right. Basically like, you know, what Mr. 
Mitchell said, anywhere from 6-9 months to come up with that 
system.
    Mr. Hall. Can you tell us more about the knowledge library? 
Though you did not mention it in your testimony, what would its 
use be once the exams are stored?
    Dr. Shahani. Basically the knowledge library that we are 
talking about or the knowledge database contains, you know, the 
rating codes that are in 38 CFR part 4. The claim conditions 
that we have encountered throughout the 9 years, and all the 
potentially claim conditions, and all other keywords that are 
within the rating code, because each rating diagnostic code 
actually has descriptions. So they will say range of motion 
limited by 30 degrees or 40 degrees. All of that is within that 
knowledge database.
    So when it actually scans the records, it identifies and 
highlights those key words. And then later links, through the 
rule-based technology or artificial intelligence, links that 
medical evidence to the rating code.
    So what the rating specialist will see is actually medical 
evidence already showing them what rating codes they need to 
consider. But they need to make the final decision.
    Mr. Hall. Thank you, Doctor. Mr. Hunter, I want to ask you 
if you could describe a little bit more about the system that 
you developed for the Navy, that you referred to during your 
testimony.
    Mr. Hunter. Well we took the approach that we have--we have 
developed what has been discussed today would take another 6-9 
months. Over the course of last 5 years, we used open standard 
technologies, realizing that no matter what software tool we 
developed would have to work with a multiple set of disparate 
systems and communicate with that data.
    So we--when you put yourself in the seat of the veteran and 
as the bureaucracy, the IUI will reflect the Boolean logic that 
is the knowledge library. It actually is that the pages refresh 
reflecting that knowledge library to say okay, based on what 
you have told us, this is the paperwork, what you need to do. 
Maybe the medical records that need to be attached. And we can 
do it without programmers. That was a real key, because you 
don't want to get--when we worked in the Navy with PeopleSoft 
and you had to write hard code APIs, you get bogged down. And 
we just do it now in drop-down menus so you can select from a 
drop-down menu that knowledge library in which to inject into 
the question.
    So we have accelerated. And the technology is not unique or 
patentable. It is just the way we presented that tool in order 
to have the institutional knowledge get transferred into that 
IUI. That is what is critical.
    Virginia refers to it as the unlucky or lucky vet. It does 
take 3-5 years to scale up a Federal adjudicator or veterans 
service representative. So if the veteran is lucky to call in 
and get someone who has just been there 2 months, well they 
don't know all the questions to ask or the right questions. And 
that is the key.
    If we don't catch this problem at the tip of the spear, it 
just rolls through the entire system. And that is what we are 
finding. They want to get the person who has had 35 years. It 
is just critical that they have a consistency of every veteran 
with the access to be asked the right questions and all the 
questions, because that starts the claim bill process.
    So some of the other companies here today can take a claims 
management process forward. And we have done that unique and 
successful--successfully.
    Mr. Hall. So how long should it take to rate a claim in 
your opinion?
    Mr. Hunter. Well in our opinion it should take less than 48 
hours, depending if the right Boolean logic is put in.
    Now I also agree it will never take the place of a human. 
All we are doing is shifting the job focus from those people 
from this data entry and doing what they really are set out to 
do, which is the human interaction.
    I also believe that the technology will never solve 100 
percent of the problems. What we found is it is more of an 80-
20, 90-10 rule. That you do not want to take the time or the 
money to put in this logic for the person with the extreme 
case. That person needs to be immediately put to personal 
attention, because they need that.
    But for the bulk of the people, the frustration of going 
through the same questions and same paper, it is ridiculous.
    And we take a position it is more about the data than the 
document. You need some documents by mandate. But documents to 
us are online receipts, box and lines around the data. It is 
the data that is really the back-end systems need, which was to 
close. We put that in standard, native XML so we can very 
confidently talk with any back-end system without trying to 
change that system. That just seemed--that really lengthens the 
time.
    They just need good data to do what they do well.
    Mr. Hall. Thank you. Mr. McGarry, I just wanted to ask 
you--well first of all, thank you for being here again. You 
have been a help to the VA system in the past. And I appreciate 
you being here again.
    It is curious, you mentioned that you can process some 
claims within 3 days but must process them within 45 days in 
order to be ERISA compliant. Should VA be required to meet the 
same standard for processing a claim?
    Mr. McGarry. I think it is certainly possible for the VA to 
meet the same standard for processing a claim. You know, my 
view of it is that the processing part isn't the only piece. 
There is the definition of disability as well as the resources 
applied.
    And so my only recommendation is in addressing this 
problem. You address all three of them to get a holistic and 
consistent solution to it.
    Mr. Hall. Does----
    Mr. McGarry. Mandating one or the other I think is going to 
be--fall short of the total solution.
    Mr. Hall. Right. Does Unum have a backlog of claims?
    Mr. McGarry. We do not.
    Mr. Hall. Do you see a lot of fraud?
    Mr. McGarry. We see--you know, fraud is a high standard 
requiring intent. We do see misrepresentation or people----
    Mr. Hall. Misunderstanding?
    Mr. McGarry. Misunderstanding. And so there is a reasonable 
amount of that. You know, we discover a fair amount of claims 
through investigation and surveillance for instance.
    Mr. Hall. Do you think that if the VA used a triage system 
similar to Unum's where the claims got sent to a subject matter 
specialist, it would improve their success?
    Mr. McGarry. Our actual triaging is less around subject 
matter specialists and more around the duration and complexity 
of the claim.
    I think one of the biggest drivers of our success is 
quickly separating claims into those that can be solved readily 
and quickly versus those that need more in-depth analysis and 
investigation.
    Mr. Hall. The 80-20 or 90-10.
    Mr. McGarry. And the thing is, you know, is don't mingle 
those two. Don't have the same people doing the 90 and the 10, 
because the 90 are quick hits that you can do in 10-15 minutes 
all day long. It is a processing work. Whereas the 10 is more 
of an investigative work that takes real expertise to do. And 
so one of our successes is separating those right up front.
    Mr. Hunter. And, Mr. Hall, may I add that quickly, we have 
found in Virginia's pilot that less than 95 percent of the 
claims are ready to rate when they are submitted by the 
hardworking VSOs in the State, only 5 percent. So if they are 
not ready to rate, the claims management process can't proceed 
properly.
    Mr. Hall. Thank you. We are going to have--since Mr. 
Lamborn is not here, we will have the Minority Counsel ask a 
few questions. And then we will move along to the next panel.
    Mr. Lawrence. Thank you, Mr. Chairman. Many of your 
questions were similar to Ranking Member Lamborn's, so I just 
have a couple.
    For Mr. McGarry and Dr. Shahani, in your testimony you 
mentioned that your systems have the capability of managing and 
organizing multiple documents.
    Veterans claims files, as you know, can be rather 
voluminous. They can submit anything they feel is pertinent as 
evidence. Would that be problematic to your systems?
    Dr. Shahani. When we ran the prototype basically and 
scanned c-files, we are able to separate duplicates. We are 
also able to separate non-medical from medical records. And so 
we don't see that to be a problem. We can build in rules again 
to separate out all those different records.
    Mr. McGarry. We have files too that stand 6 feet tall 
stacked one on top of the other, which is why it is such a must 
to have a document--a document management system is such a big 
piece of the file so that you can footnote and identify those 
documents that are germane to the decision.
    Mr. Lawrence. Thank you. And, Mr. McGarry, how long did it 
take Unum to establish your system?
    Mr. McGarry. It took approximately 3 years.
    Mr. Lawrence. And for Mr. Mitchell, you had mentioned 
additional benefits that may accrue from more advanced 
technologies that can be adopted once the claims are captured 
and managed online. Could you elaborate on that just briefly 
please?
    Mr. Mitchell. Sure. I was primarily thinking of data mining 
that collection of benefits claims and how they were ruled on 
finally. So if you had that kind of data, you could data mine 
that for example to detect the features of the claim that 
indicate, for example, that this is likely to require a 
particular type of special processing. And to do the kind of, 
you know, initial sorting that these gentleman were talking 
about.
    You could do data mining to detect the features of the 
claim that suggest perhaps this should be looked at as a 
potential case of fraud or misunderstanding. That is very 
common in the insurance industry.
    So primarily I was thinking of the--of the uses of that 
data in a data mining.
    Mr. Lawrence. Thank you.
    Mr. Hall. I would like to thank our panel. It is very, very 
interesting. You have exceeded my expectations. I don't know 
about anybody else, but I trust that these are very exciting 
possibilities that you raise. So thank you again for your 
testimony, and for your responses to our questions.
    This panel is dismissed, And we will ask our third panel to 
come forward.
    Kim Graves, the Director of the Office of Business Process 
Integration of the Veterans Benefits Administration, U.S. 
Department of Veterans Affairs, and Stephen W. Warren, 
Principle Deputy Assistant Secretary of the Office of 
Information and Technology, U.S. Department of Veterans 
Affairs.
    And if we are lucky, the votes will be held off until after 
we hear from our two panelists and ask a couple of questions.
    While you are getting settled, I will tell you that within 
the last couple of months, our office up in New York's 19th 
District resolved a claim for a Navy vet from World War II 
which was the most extreme case that I have come across yet.
    A man who had two ships blown out from under him in the 
Pacific, one by a kamikaze pilot, one by a torpedo. Twice was 
floating in the ocean with sharks and body parts floating by 
him. Had to be pulled back off the ship by his buddies, because 
he kept on trying to rescue more of his shipmates and get them 
in the lifeboat.
    He has a drawer full of medals for it. He is 84 years old 
now, and had been diagnosed as schizophrenic, which of course 
is not a service-related diagnosis. With the help of his friend 
who happens to be the local Veterans of Foreign Wars commander 
of the post that he belongs to, and my staff, and working with 
the local VA, and the VSOs in our area, and so on, we corrected 
it.
    And Sailor Ken McDonald had a happy Christmas with $100,000 
of back disability pay, and $2,400 a month, and 100 percent 
Post Traumatic Stress Disorder rating, which is evident when 
one talks to him about--even all these years after. He was 20 
when these incidents happened. Yet today he still shakes and 
has a hard time, when you bring it up and ask him about it.
    But we can prevent worst case scenarios. I guess the worst 
case is, if he didn't live to have the resolution. But 
hopefully we can move this all toward a quicker, more efficient 
resolution.
    And Director Graves, we have your statement--your written 
statement is in the record. So you have 5 minutes give or take. 
And you are now recognized.

   STATEMENTS OF KIM A. GRAVES, DIRECTOR, OFFICE OF BUSINESS 
  PROCESS INTEGRATION, VETERANS BENEFITS ADMINISTRATION, U.S. 
    DEPARTMENT OF VETERANS AFFAIRS; AND STEPHEN W. WARREN, 
   PRINCIPAL DEPUTY ASSISTANT SECRETARY FOR INFORMATION AND 
    TECHNOLOGY, OFFICE OF INFORMATION AND TECHNOLOGY, U.S. 
                 DEPARTMENT OF VETERANS AFFAIRS

                   STATEMENT OF KIM A. GRAVES

    Ms. Graves. Mr. Chairman and Members of the Subcommittee, 
it is a privilege to be here today to talk about the use of 
information technology to enhance claims processing within the 
Veterans Benefits Administration.
    VBA has made significant strides in the use of information 
technology to improve claims processing in all of our benefit 
programs.
    Our current focus is the development of a comprehensive 
strategy to integrate the various initiatives already underway 
and leveraging successes already accomplished. VBA is 
collaborating with the Office of Information and Technology in 
developing this strategy to ensure our mission needs are met 
and that the appropriate enterprise architecture is employed.
    At the core of our strategy is the implementation of a 
business model for compensation and pension processing that is 
less reliant on paper documents. The use of imaging technology 
and computable data to support claims processing in our 
insurance, education and loan guaranty programs has been 
successful for many years.
    Initial pilot efforts in our compensation and pension 
business line have demonstrated the feasibility of using this 
type of technology for these benefit programs as well.
    Our comprehensive strategy, the Paperless Delivery of 
Veterans Benefits initiative, is envisioned to employ a variety 
of enhanced technologies to support end-to-end claims 
processing.
    In addition to imaging and computable data, we will also 
incorporate enhanced electronic workflow capabilities, 
enterprise content, and correspondence management services, and 
integration with our modernized payment system, VETSNET. In 
addition, we are also exploring the utility of business rules 
engine software for workflow--for both workflow management and 
to potentially support improved decisionmaking by claims 
processing personnel. A recent request for information (RFI) to 
industry yielded a variety of products that may be useful in 
our end-state vision.
    As part of our strategy for improving the claims processing 
business model, VBA recently contracted with IBM to conduct a 
study of the current process and suggest improvements. We 
expect their report shortly and will assess their findings as 
we move forward with documenting our information technology 
strategy.
    As noted previously, two pilot programs are currently 
underway and have demonstrated the utility of imaging 
technology in our compensation and pension business line. Both 
projects utilize our Virtual VA imaging platform and related 
applications. Virtual VA is a document and electronic claims 
folder repository.
    The first pilot supports our income-based pension program. 
It involves imaging documents received in conjunction with the 
annual income reporting process.
    Imaging allows the three Pension Maintenance Centers to 
make the necessary claims adjustments without need for 
retrieval and review of the paper claims file.
    The second pilot supports the compensation program at our 
centralized rating activity sites for our Benefits Delivery at 
Discharge program. The separating servicemember's medical 
records and supporting claim information are imaged at the 
outset of the claims process. This allows rating veteran 
service representatives to make decisions based solely upon 
review of the imaged records rather than reliance on the paper 
claims file.
    Further refinements of the business process are now 
underway and will be factored in as we evaluate options for 
expanding use of this technology.
    An additional pilot project is also under development. This 
project will examine issues such as user authentication and 
using online forms to provide the capability for the initial 
``electronic'' filing of benefit claims. This is a first step 
in implementing online self-service to allow veterans to manage 
some of their interactions with VA electronically.
    Integration with VETSNET is also a critical success factor 
in our overall strategy. We have made significant progress in 
the implementation of VETSNET over the past 2 years.
    Approximately 98 percent of all original compensation 
claims are now being processed end-to-end in VETSNET. And we 
are now paying monthly compensation benefits to more than 
850,000 veterans or approximately one of every three 
compensation recipients using our modernized platform.
    With our next conversion of records from the legacy 
Benefits Delivery Network scheduled for April, VETSNET will 
become the primary payment system for compensation benefits.
    Integration and data exchange with the Department of 
Defense are also essential, as is our continued expansion of 
exchange of healthcare information with the Veterans Health 
Administration.
    As we continue to move forward with the efforts described 
here, we are focused on developing an integrated project plan, 
ensuring the needs of our veterans and their families are 
documented and attainable. Demonstrable milestones and 
performance metrics will be incorporated so that we and our 
stakeholders are able to assess our progress in achieving our 
vision.
    To assist in developing this plan, we are working closely 
with our Office of Information and Technology partners to 
develop a request for proposals to engage the services of a 
lead systems integration contractor.
    The integrator will provide support in documenting both the 
business and technical requirements for implementation of our 
long-term strategy.
    I assure you that the Under Secretary for Benefits is 
committed to implementation of the Paperless Delivery of 
Veterans Benefits initiative.
    Together with our partners in the Office of Information 
Technology, we believe this goal is not only attainable, but is 
imperative to ensure the best possible service to our Nation's 
veterans.
    We thank you for the opportunity to address these important 
issues and would be happy to address any questions that you may 
have. Thank you, Mr. Chairman.
    [The prepared statement of Ms. Graves appears on p. 78.]
    Mr. Hall. Thank you, Ms. Graves.
    Mr. Warren, you are now recognized for 5 minutes.

                 STATEMENT OF STEPHEN W. WARREN

    Mr. Warren. Mr. Chairman and Members of the Subcommittee, I 
would like to thank you for the opportunity to testify today on 
the use of information technology to enhance claims processing, 
within the Department of Veterans Affairs, as well as utilize 
data from the Veterans Health Information Systems and 
Technology Architecture or VistA system to assist in the 
processing of disability claims. These are very important 
issues that affect the life of every veteran and their just 
compensation for disabling injuries received while serving our 
Country.
    I would like to begin by addressing VA's efforts at 
leveraging information technology to improve the timely 
delivery of veterans' benefits. The Office of Information and 
Technology has been collaborating with the Veterans Benefits 
Administration in the development of a comprehensive strategy 
to achieve their target business model.
    The operational concept of the Paperless Delivery of 
Veterans Benefits initiative is to employ enhanced technology 
platforms to include imaging, computable data, electronic 
workflow capabilities, and enterprise content and 
correspondence management services. Some of the same 
technologies you heard from earlier panelists.
    The initiative will integrate with the Veterans Benefits 
Administration's core business application and modernized 
payment system, the Veterans Service Network known as VETSNET.
    My office also supports the Veterans Benefits 
Administration's market research of business rules engine 
software and other decision support technologies, which can be 
leveraged to improve and expedite decisionmaking by claims 
processing personnel.
    We recently released a Request for Information from 
industry or RFI. This request for information resulted in the 
demonstration of technologies that may be appropriate for the 
Veterans Benefits Administration's target business strategy.
    The request for information process helps us gain a better 
understanding of how private industry and other government 
agencies have employed these types of technologies to support 
their specific business models.
    We also have conducted an analysis of technical 
architectures, business applications, and Commercial Off-The-
Shelf products, utilized to support the business processes of 
the Social Security Administration, as well as the Veterans 
Affairs Organization of Australia and Canada.
    A Statement of Work is currently being prepared to engage 
the services of a Lead Systems Integration Contractor. The 
purpose of this contract is to assist with the development of 
the overarching strategy and business requirements for the 
Paperless Delivery of Veterans Benefits initiative.
    These key deliverables will enable us to begin specifying 
the supporting technical architecture and business 
applications.
    Mr. Chairman, I would like--now to highlight how the 
utilization of data from the VistA system, the one used by the 
Veterans Health Administration, assists in the processing of 
disability claims.
    The business application used by the Veterans Benefits 
Administration to navigate and retrieve clinical data within 
the VistA system, is called the Compensation and Pension Record 
Interchange or CAPRI. Online access to medical data, housed in 
the Veterans Health Administration VistA system, supports the 
disability benefits determination.
    CAPRI also provides access to some Department of Defense 
medical records through integration with the Federal Health 
Information Exchange framework. CAPRI was nationally deployed 
during fiscal year 2001, and delivered cutting edge ``point and 
click'' technology to the users' desktop at that time.
    Since its deployment, the application has been repeatedly 
enhanced as new categories of clinical data in the Veterans 
Health Administration and Department of Defense became 
available.
    Mr. Chairman, in closing I want to assure you that we 
remain steadfast in our efforts to continuously optimize any 
and all information technology improvements, as we strive to 
improve our veterans' benefits IT environment.
    Our goal is that these efforts, coupled with the support of 
the Veterans Benefits Administration and our partners in the 
private sector, will greatly improve the business processes, 
which will significantly enhance the disability claims process 
that our Nation's heroes undergo.
    Thank you for your time and opportunity to address these 
issues. I would be happy to answer any questions you may have.
    [The prepared statement of Mr. Warren appears on p. 79.]
    Mr. Hall. Thank you, sir.
    Ms. Graves, I just wanted to ask you, in November of 2007, 
VA testified that it has received $20 million in a supplemental 
appropriation for electronic processing initiatives.
    When this Subcommittee asked about IT expenses, we were 
given an analysis that showed that VA spent approximately $300 
million on VETSNET since 1986. With all that time and money, 
how is it that we still do not have a system that satisfies 
veterans' claims processing needs?
    Ms. Graves. Thank you, Mr. Chairman. Moving to the VETSNET 
environment, off of our antiquated and outdated payment system 
has been a paramount concern to the Veterans Benefits 
Administration.
    The actual software development component of VETSNET began 
in 1996. And it has taken us a significant amount of time to 
make these accomplishments.
    Over the past 2 years, Under Secretary Cooper instituted a 
variety of changes to include restructuring the overall 
management of the VETSNET project. We believe that the progress 
that we have made demonstrates that we have learned some very 
significant lessons in how to better manage the business 
process of IT development.
    We hope to bring to bear these lessons learned as we move 
forward with our next initiative, which is that next step in 
bringing a paperless environment to the Veterans Benefits 
Administration.
    Mr. Hall. It sounds like your current plan to make the 
system paperless means that the rater does the same things with 
the screen that they did with the paper record unless I am 
mistaken.
    What is the plan to make an electronic record computable so 
that data can be mined, matched, and manipulated?
    Ms. Graves. There are a number of efforts that we will have 
to address with our information technology partners.
    I think as you heard in the prior panel, some of the issues 
that we must contend with deal with the records of veterans who 
may have exited the service many years ago. Many of these 
documents are handwritten.
    This presents a number of challenges in turning that into 
computable data. We will be looking at all of our opportunities 
for moving forward from a paper environment, whether it is 
images, computable data, all along the spectrum, to enable us 
to better utilize the data in support of claims processing.
    I think as you heard from the panels before, the rating 
process in and of itself is significantly difficult. There is 
much human judgment that must be applied. It is not only a 
matter of determining a level of disability, but making a 
judgment as to whether the disability itself was incurred in or 
aggravated by service.
    So there are a number of factors that must be brought to 
bear. And as we work with our IT partners, we will be looking 
for all of the opportunities that we can utilize to facilitate 
bringing this to a more streamlined process.
    As we mentioned in the testimony, IBM Global Services has 
been with us for the past number of months conducting a study 
of the claims process. We are anxiously awaiting their findings 
to help us look at the business model itself and match that up 
with technologies that are available to improve the claims 
process.
    Mr. Hall. Thank you. The VA already has Veterans Online 
Application (VONAPP) so that veterans can file online. So, I am 
wondering what would be the purpose of an additional pilot 
project you mentioned to study this capability if the capacity 
already exists, or are they dissimilar?
    Ms. Graves. The current VONAPP process that we have is an 
online application. The veteran can fill in the application, by 
typing in their information in the application. They can either 
email that document, that application into us, or mail it in 
hard copy.
    In either case, we accept the application. We must also go 
out and get a physical signature from the veteran. Regardless 
of whether they have submitted it online, we must have the 
signature.
    And also in its current form, the information that we 
receive on the veteran's application is re-keyed into the 
claims processing system.
    The pilot that we are working with our IT partners on will 
take the next step, and hopefully begin to utilize fillable 
forms, computable data, and also explore our ability to accept 
an electronic signature as we move forward with the appropriate 
business process that will allow us to accept that online 
signature.
    Mr. Hall. Thank you, Ms. Graves. Mr. Warren, what is the 
Under Secretary for Benefits' commitment to fully instituting 
an automatic claims processing system? What has he done in 6 
years, in your opinion, to get to this goal, and why has it 
taken so long with so few results?
    Mr. Warren. Mr. Chairman, I probably should caveat my 
remarks with stating that I have only been at the VA for the 
last 9 months. So I can only give you my observations for the 
last 9 months.
    Mr. Hall. That is good.
    Mr. Warren. The commitment that I have seen by our partners 
in the Veterans Benefits Administration is a commitment to make 
the dramatic changes necessary to go forward.
    And I think one of the things that it is good to keep in 
mind as we talk about how do you take an old paper-based system 
and move it into the nirvana, if you will, or at the punch of a 
button it makes a determination, there are many steps you need 
to go through. And some of them were touched upon.
    It is moving from paper data to electronic data. It is 
moving to electronic data that is computable. It is utilizing 
workflow tools or technologies that allow you to move the 
information to the appropriate folks.
    Then there is the need for tools that assist in the 
determinations up to the point where maybe you can have a 
tentative determination. And then somebody having to look at 
it. Each one of those things take time, especially with a 
consideration for what are the rules that the organization has 
to follow?
    And the Department, through the Veterans Benefits 
Administration, is looking at those rules and trying to 
understand what does it take to automate those rules and are 
there limitations in the rules themselves?
    As an example, our colleagues in the Australian Veterans 
Administration went through this process themselves. And 
working with their legislature, it took them 4 years from going 
to ``we want to do this'' to, ``how do we need to change the 
rules?'' How do we need to make the rules actionable, so we 
actually can use automated tools to make the determinations?
    So I would love as a veteran for it to be easy. However, we 
have complex rules and complex systems that need to be taken 
forward through a deliberative process so we don't mess it up 
along the way, sir.
    Mr. Hall. Some of our previous panelists suggested, coming 
from their private sector positions, their view was that a 
system like this could be developed in 6-9 months. Do you think 
that is accurate? I don't mean completely dialed in and have 
all the data entered into it, but to have the actual system.
    Mr. Warren. And to give you a sense in terms of how quickly 
you can do things, prior to the Department of Veterans Affairs, 
I was the Chief Information Officer at the Federal Trade 
Commission. And we brought on the Do Not Call Registry in 100 
days. So you can do complex things quickly. But the National Do 
Not Call Registry is actually a trivial effort in comparison to 
what it will take to make the system, and the processes, and 
the rules that the folks need to use into an automated system.
    I wish it was 6-9 months, because then we could get it 
done. But it actually is going to take longer once you look at 
the complexity of the rules and the ambiguity in some of the 
rules. And we will need your assistance and the assistance of 
this body as we identify what rules might be too ambiguous for 
the utilization of advanced technology to make the 
determination.
    Mr. Hall. There used to be a VA Office of Seamless 
Transition, which has now become a VHA/DoD Outreach 
Coordination Office. How has VBA been dropped from the process? 
How are veterans, who seem to have a difficult enough time in 
getting claims processed, supposed to navigate the system 
without this level of support? I guess that could be to either 
of you. But----
    Ms. Graves. Mr. Chairman, I apologize. But I am not aware 
of the change that you have stated.
    Mr. Hall. Referred to?
    Ms. Graves. Referred to. We can certainly take that 
question back and get you a response on that. I apologize for 
not having that information.
    [The response was provided by VA in the answer to Question 
3 from the post hearing questions for the record, which appears 
on p. 101.]
    Mr. Hall. Well, I will just ask one more. And then turn it 
over to Mr. Lamborn for his questions. But I am just wondering 
why the Clevelands, Gunnery Sergeant and Mrs. Cleveland, even 
though they went through the BDD process and--well, you--were 
you here for their testimony?
    Ms. Graves. Yes, sir.
    Mr. Hall. Do you--would you hazard a guess as to why it 
would take a year to rate and compensate Mr. Cleveland? Or is 
this just one of those stories that you hear about where things 
fell through the cracks repeatedly?
    Ms. Graves. Mr. Chairman, first and foremost, I want to 
apologize and did apologize to Mr. and Mrs. Cleveland for the 
difficulty that they endured moving through the claims process.
    I do not have the specifics on the timeline that Mrs. 
Cleveland so eloquently went through. I give you my assurance 
that we will be doing that when we go back to the office to 
make sure that we have not only addressed any issues that still 
may be outstanding. Unfortunately, when we become aware of 
cases that in falling through the cracks is such a--it doesn't 
do justice obviously to what the family endured.
    But certainly as we become aware of these types of 
circumstances, we do look at these. And try to make adjustments 
where we can to ensure that we put procedures in place to try 
to prevent these from occurring in the future.
    Mr. Hall. I would guess that maybe we were talking about 
the 80-20 or 90-10. And this might be in the 10 or the 20. In 
other words, it is the more dramatic instances of evident 
disability. Like I have had a couple that my office has dealt 
with like the sailor with the two ships blown out from under 
him for instance. If one had 80 or 90 percent of the cases 
being processed with the computerized, automated system 
primarily, and then have the ones that need special care being 
diverted to human resources, you would hope that would solve 
the problem.
    I am just curious it would seem that an automated registry, 
a record, right from the word go, would eliminate the many 
times they were asked to resubmit, the many times they were 
told the record couldn't be found, that there was no ``this or 
that'' form or medical report.
    So let us all hope that we are after the holy grail here. I 
guess my last question to either or both of you is are the 
systems that you already have in place and that you are--that 
the VA is developing, going to be, or are the people who have 
put those in place going to be, open to changing them or adding 
things like some of the previous panelists talked about?
    Ms. Graves. If I may, certainly one of the lessons that we 
have learned in the last couple of years with the VETSNET 
initiative that there is a pull from our employees for the 
types of technologies that we have been delivering and that we 
will continue to deliver.
    Our paperless rating process and the benefits delivery at 
discharge pilot has also demonstrated from our ratings 
specialist, our rating veterans service representatives, that 
not only can they rate a claim in a paperless environment, but 
they prefer it, at least the ones who have gone through that 
process.
    That has given us a demonstrated capability that this is 
something that can be accepted and will be accepted by at least 
a group of our employees.
    So we are looking at that process right now on how we can 
expand that. And that would certainly, in taking Mr. 
Cleveland's case, if, as we expand the BDD process and the 
paperless BDD process, when Mr. Cleveland--coming through now, 
were in that population, we would have received his paper 
records at the time of his discharge. And then imaged them 
immediately into the system. So at least that opportunity for a 
loss of a record would be certainly greatly mitigated or 
diminished.
    So we believe that our employees are open. And would 
welcome the advanced technologies that are available. And we 
are very anxious to set a course that is achievable and to move 
forward to a better system for our veterans.
    Mr. Hall. Thank you very much, Ms. Graves.
    The Chair will now recognize Ranking Member Lamborn.
    Mr. Lamborn. Thank you, Mr. Chairman. And this is for 
either one of you. Can you give some examples of some of the 
types of questions that you expect the IBM study to be able to 
answer?
    Ms. Graves. Thank you. We have just received a very 
preliminary report from IBM. But we are hopeful that the IBM 
group will be able to point us in a direction of where we may 
be able to improve the process.
    I certainly expect that we will hear from IBM some of the 
things that we have heard today on the panel. That our reliance 
on a paper-based system is detrimental to the overall 
efficiency of the process. I am speculating on that. But 
certainly we would expect to derive great benefit from IBM's 
observations as they have gone and looked at our claims 
process.
    Mr. Lamborn. Now apart from the rules-making ability or 
excuse me the claims adjudicating ability that hopefully will 
eventually be realized as a goal. Just in the meantime, it 
seems that document management would be a huge benefit. You 
know seeing the picture that the Chairman showed of an eight-
inch stack of paper, or we saw something like that on the table 
in front of us today.
    Is the IBM study looking at that only, or are they going 
beyond that? Or what are they looking at again?
    Ms. Graves. The IBM study was designed as a comprehensive 
review of the compensation claims process. I believe their 
charge was to come back with any suggestions that they would 
have, whether it is regulatory, legislative, or information 
technology that might be brought to bear to improve the claims 
process.
    The pictures that were shown today and the look, the 
physical look at Mr. Cleveland's--the portion of Mr. 
Cleveland's records, again, certainly demonstrates that as we 
become better able to turn that paper into something that is 
easier to manager, easier to keep control of, we can only 
speculate that will improve our ability to manage that 
workflow.
    Some of the technologies that we have been exploring with 
our partners in the Office of Information and Technology, we 
are looking forward to evaluating how those types of workflow 
management and document management tools will enable us to not 
only take those paper records and turn them into an image or 
some type of computable data, but also to manage the flow of 
that information throughout the claims process.
    Mr. Lamborn. Now I take it that they are not looking at the 
ability to come up with decisions through artificial 
intelligence or anything like that. They are not going that far 
on the cutting edge, are they?
    Ms. Graves. Sir, I would presume that if IBM, in their 
review of the process, believes that is a viable opportunity 
that they will present that. Their charge was relatively open 
to come in and review the claims process and provide 
recommendations for improvement. And they were not constrained 
on the types of the improvements that they can provide to us.
    Mr. Lamborn. And you have seen a preliminary version of 
that?
    Ms. Graves. Very preliminary. Just a couple of pieces of 
it. I have not had a chance to go through it. And it has not 
been formally released to the Veterans Benefits Administration 
yet.
    The only look I got at it was to ask so they could ask a 
couple of clarifying questions as they were putting some 
touches on their draft.
    Mr. Lamborn. Okay. Well I look forward to helping or 
learning with you some of their recommendations and helping the 
VA as we go forward to make this a better process. Whether it 
is just document management or even beyond that into the 
processing of claims.
    So thank you for what you are doing. And thank you for your 
testimony today.
    I yield back, Mr. Chairman.
    Mr. Hall. Thank you Mr. Lamborn. Ms. Graves and Mr. Warren, 
thank you for your testimony and for the work that you are 
doing. And I echo the Ranking Member's comments that we are 
here to help, and to nudge, and to stir the pot.
    And we would love to see a copy of the report as soon as it 
is in a presentable enough form that you can share it with us. 
The sooner the better. It couldn't happen too fast for us.
    We all have veterans in our own districts that we deal with 
on a day-to-day basis and our staffs deal with on a day-to-day 
basis. And we see, as with the Clevelands here, the cases I 
think that get to a Congressional office are the ones that have 
had trouble. So we don't necessarily see a scientific sample.
    But what we see are the ones where the system failed to 
come through in an adequate or in a timely fashion. And that 
is, for me, what motivates me, and I think all of us to, want 
to cover and take care of our veterans without these problems 
arising, and to give them the service that is commensurate with 
that they gave to our country.
    So thank you for your comments. Thank you for your 
conversation with the Clevelands', which I also had. I think 
that it is the best thing that we can do as a tribute to them 
and to others like them is to continue and speed up this 
process of modernizing a system that, as one of the previous 
witnesses said, we would not tolerate if it were our own health 
insurance.
    In private business we have grown accustomed to a higher 
standard or quicker standard of technological resolution of 
these issues.
    So institutional momentum being what it is, we are going to 
work together and move into the 21st century with both feet.
    So if there are no further statements, no further 
questions, I thank you and all the panels. And this hearing 
stands adjourned.
    [Whereupon, at 4:54 p.m., the Subcommittee was adjourned.]





















                            A P P E N D I X

                              ----------                              

           Prepared Statement of Hon. John J. Hall, Chairman,
       Subcommittee on Disability Assistance and Memorial Affairs
    I would ask everyone to rise for the Pledge of Allegiance--flags 
are located in the front and in the rear of the room.
    I would first like to thank the witnesses for coming today to 
appear before the Subcommittee. I know I speak for my colleagues when I 
say that we are all extremely frustrated and disappointed when we hear 
about 650,000 claims pending and another 147,000 appeals with a delay 
of 183 days to process those claims. But looking at this photograph of 
an 8 inch paper record held together with rubber bands and marked with 
post-it notes, it's hard to imagine that things don't get lost or 
missed. This has got to be cumbersome to process.
    There is no doubt that we need a better system than rubber bands 
and post-it notes and must look beyond the current way VA is doing 
business. There are best practices within the scientific community and 
in use in the private sector. I thank you for joining me today to 
explore those solutions and to broaden our understanding of what is 
possible, realistic, and achievable in this technological age.
    The current VA claims process is paper intensive, complex to 
understand, difficult to manage, and takes years to learn. Training a 
rater can take 2-3 years and many leave within 5 years. Experienced 
raters can adjudicate about three claims a day taking about 2-3 hours 
apiece. This means that if there are 10 people who can rate a claim and 
800 claims are ready to rate, then it will take another 80 days to 
process those pending claims, which have already been in the system for 
several months. This is very labor intensive and in the meantime 
veterans are waiting months without compensation while their completed 
case sits on a shelf. I find that unacceptable.
    Additionally, there have been reports by the GAO, the VA Inspector 
General, and the Institute for Defense Analyses (IDA) that explored the 
variances in ratings between the Regional Offices and the lack of 
inter-rater reliability. The Veterans' Disability Benefits Commission 
also found a great deal of subjectivity and inconsistency in the VA 
disability claims process.
    So how do we solve this?
    I've had a lifelong interest in science and was a three-time 
National Science Foundation scholar in High School and studied physics 
at Notre Dame. So, I find the topic of Artificial Intelligence--or AI--
compelling since it requires the confluence of science, mathematics, 
engineering, and physics. In general, the purpose of AI is to make 
computer programs--or machines--that can solve problems and achieve 
goals. AI software increases speed, improves accuracy, and reduces 
costs for many industries and agencies. AI does not replace the human 
element, but rather facilitates its availability. There are great 
examples of AI in other areas, such as banking and medicine. For 
instance, the Veterans Health Administration relies on VistA to help 
doctors with diagnosis and treatment. It sends alerts when a patient 
needs a flu shot, cholesterol screening, or warns of potential drug 
interactions.
    AI can be a decision support tool for adjudicating claims too. It 
could be used to organize and store data. It could match key words from 
a veteran's record to the criteria in the Rating Schedule. It could 
prioritize multiple disability issues.
    I envision a VA in which a veteran can apply online for benefits, 
upload records, exams, and other certificates, which are prioritized 
and classified by an expert system that can match the data to the 
Rating Schedule criteria and shorten the time it takes to generate a 
claim. The electronic template used by the examiner could be associated 
with the Rating Schedule, which could also calculate ratings. 
Classifiers or key words could easily be matched by the computer to the 
Rating Schedule, such as if ``Arm,'' ``Amputation,'' then ``90 
percent.''
    This would free up the time for Regional Office employees to deal 
with the more complicated issues and assist veterans and their families 
with their problems. This Subcommittee has often heard that veterans 
don't know about or understand their benefits, and that transitioning 
servicemembers are not getting all of the support that they need from 
the VBA. In this way, VBA staff could be providing more outreach and 
ensuring that veterans understand their entitlements and eligibility 
requirements for other such programs as Vocational Rehabilitation, 
insurance, and special monthly compensation.
    I am eager to hear testimony today that will open up the discussion 
on information technology and share ideas that can improve rating 
efficiency, quality, and accuracy while reducing inconsistencies and 
variances in decisions for our disabled veterans who are waiting on a 
claim determination.
    I look forward to working with Ranking Member Lamborn and the 
Members of this Subcommittee in finding real solutions that will vastly 
improve the VA claims process. It is unconscionable that veterans are 
waiting as long as they are for their earned benefits and that must 
end.

                                 
  Prepared Statement of Hon. Doug Lamborn, Ranking Republican Member, 
       Subcommittee on Disability Assistance and Memorial Affairs
    Thank you Mr. Chairman for yielding. I would like to welcome all of 
our witnesses to the Subcommittee's first hearing of the second 
session.
    I want to commend you Mr. Chairman for your leadership and 
bipartisanship in the previous session and I look forward to working 
with you and your staff to find meaningful solutions to improving the 
VBA claims processing system and reducing VBA's disability claims 
backlog.
    I am excited that our topic of discussion today is the use of 
artificial intelligence to improve the disability claims process.
    As you know Mr. Chairman, this is an idea that my colleagues and I 
on this side of the aisle have long supported.
    Whether it was in our FY08 views and estimates, or two bills that I 
introduced last session, H.R. 1864 and H.R. 3047, we believe that one 
way to truly reduce the current backlog and prevent future backlogs is 
to propel VA beyond a 20th century paper-based processing system.
    VA must create a system where all claims are electronically scanned 
and rating board members have access to computerized interactive tools 
to assist them in the adjudicative process.
    Hopefully the new system will lead to more accurate rating 
decisions that are delivered to our Nation's veterans in a timely 
manner.
    While I envision an important role for artificial intelligence in 
the decisionmaking process, I also concur with our witnesses who will 
attest that this technology should not, and will not, ever completely 
replace claims adjudicators.
    A few weeks ago, staff from both sides of the aisle attended a 
briefing where VBA laid out plans to move forward with such a system 
and I am excited to learn more about those plans today.
    This Subcommittee must ensure that this new initiative is fully 
funded and completed with the speed and attentiveness that our veterans 
deserve.
    I am glad that we have representatives from both the private and 
academic sectors here with us today.
    It is my hope that they will be able to help VA develop some of the 
options that are currently available in the private sector.
    While I understand that VA has a very large and unique disability 
claim system, there are similar systems out there and I would hope that 
VA would look at these systems before they re-invent the wheel.
    We must improve this system so heroes like Gunnery Sergeant 
Cleveland do not have to wait several years to have their claim 
adjudicated correctly.
    Mr. Chairman I extend my thanks to you and your staff for holding 
this hearing this afternoon and I look forward to hearing the testimony 
of our witnesses. I yield back the balance of my time.

                                 
   Prepared Statement of Gunnery Sergeant Tai Cleveland, USMC (Ret.),
                    Dumfries, VA (Disabled Veteran)
    TAI: Mr. Chairman, Ranking Member Lamborn, distinguished Members of 
the Committee, thank you for the opportunity to testify before you 
today regarding my experiences with the Department of Veterans Affairs 
(VA) claims process. My name is Gunnery Sergeant Tai Cleveland (USMC, 
Ret.). With me today are my wife Robin and my children, Brittiney and 
Rudi. My other son, Tai Jr. could not be with us today. With your 
permission, as Robin has most often dealt with the VA on our benefits 
and claims issues, I have asked her to deliver our testimony.

    ROBIN: Thank you, Mr. Chairman. My husband served his country 
proudly for 24 years as a United States Marine, and although we had 
many issues with the Department of Defense following his injuries, due 
to the subject of the hearing, I will limit my comments to our 
difficulties with the VA claims processing system and its impact on our 
family. As I am speaking, however, please keep in mind, that a severely 
injured servicemember must navigate multiple systems, the Department of 
Defense, the Social Security Administration, Medicare and the VA. It is 
quite overwhelming to say the least.
    Tai was injured in August 2003 during a hand to hand combat 
training accident in Kuwait where he was flipped onto his back injuring 
his head and multiple vertebras. The resulting damage has left my 
husband a paraplegic with chronic neuropathic pain, spasticity and what 
is classified as a mild to moderate Traumatic Brain Injury that has its 
own set of challenges.
    Since Tai's injury I have had to learn the hard way how to navigate 
the systems; keeping meticulous records of documents, recording dates 
and times of telephone calls, and confirming receipt of anything sent 
or hand delivered to Federal agencies. As such, I thought the best way 
to convey our situation was to share a timeline detailing our 
experiences with the VA.
    In June 2005 we attended the Transition Assistance Program (TAP) 
class, provided by the Marine Corps and the VA to learn about the 
available options. We completed the VA's Benefits Delivery at Discharge 
(BDD) process--including the benefits, specially adaptive housing, and 
adaptive vehicle program applications--and hand-delivered it with 
medical records, MRI compact discs, films, prescription reports, etc., 
to 1722 I Street, Washington, DC.
    After having completed his compensation and pension exam, we called 
the VA Benefits number at 1-800-827-1000 in November of 2005 and were 
advised that the application was incomplete and medical records from 
the Military Treatment Facility were needed. I delivered a second copy 
of MTF medical records to the DC Office.
    A month later, I again phoned VA Benefits to see if the records 
were received and was advised that no application was on file. I copied 
and redelivered the original application to DC Office.
    In January 2006, I made another call to VA Benefits and was advised 
that the claim was being reviewed but that medical records were 
required to make a final determination. I AGAIN copied medical records 
and redelivered to DC Office. I was later told that the housing and 
vehicle grant were denied.
    When I called in February of 2006, I was told that no determination 
could be made because Tai was still on active duty. Additionally, I was 
told that no claim was on file for housing or vehicle which are allowed 
while on active duty. We reapplied.
    In March of 2006, I met with a VA employee at Walter Reed regarding 
benefits and our difficulty with the claim. She introduced us to a VA 
social worker at Walter Reed who enrolled Tai in the Adaptive Driving 
Program at the Richmond VA. We were told to reapply for benefits 
because no applications were found. We resubmitted the original 
application and completed a new application for Specially Adaptive 
Housing, HISA, and Vehicle Grant, but were informed on April 5 that the 
applications were denied and advised to reapply yet again.
    In June 2006 we were informed by the VA social worker that approval 
for the vehicle application was received but she was, ``unable to 
locate our application because the clerk failed to separate application 
and keep an in-house copy.'' In addition our HISA and Adaptive housing 
grants were denied. We reapplied.
    Everything was quiet for the next 3 months until October 26, 2006 
when we were advised to reapply for vehicle/housing grants since no 
official notification of approval was received.
    Again in November of 2006 we received verbal notification from the 
VA Rep at WRAMC of the latest vehicle and housing denial, and on 
December 13, 2006, we were advised to reapply for vehicle and housing 
grants and were contacted by VA to verify our address.
    In January of 2007, Tai was medically retired from the Marine Corps 
and after filing BDD, we assumed we would get his disability check 
within a month or so.
    In February of 2007, our housing and vehicle grants were approved, 
and had supposedly been approved since April of 2006, but the hardcopy 
was no longer on file. To date, we have still not received an official 
vehicle approval.
    In late May 2007 we received verbal notification from the VSO 
helping us at the time that the VA was indicating that there was not 
enough information on file to rate the claim, and, therefore, 
additional information was necessary.
    In June, we received notification from the VA of an 80% partial 
rating. We were advised that the rating was temporary and additional 
information was necessary in order to process the claim. As we were 
scheduled to be in Richmond shortly to obtain an adaptive cycle, we 
were advised to have Richmond perform the necessary evaluation for 
submittal to the Roanoke Regional Office. While at Richmond, I also 
inquired about obtaining a vehicle grant hardcopy and contacted the VA 
to inquire about Aid & Attendance. I was told that I was not eligible.
    In July 2007 we delivered, via express mail, Tai's medical records 
from Richmond to Roanoke and sent the VA an email advising that we 
still had not received a disability check approximately 6 months post-
discharge.
    In August I phoned and emailed VA Benefits again and told them that 
despite the temporary rating, we still had no check. I requested direct 
deposit information and requested to verify our address.
    After having been contacted about our problems by a non-profit 
organization, a concerned representative from the VA's Central Office 
called in September about the outstanding checks and we were told that 
a tracer would have to be placed on the missing checks before 
replacements could be issued. I later received a call from the Roanoke 
office and was advised that replacement checks were going to be issued.
    On October 4, 2007, a VA Representative told us that claim was 
being expedited and should be completed by the 14th. We were informed 
on the 14th and the 30th that the updated medical report still had not 
been received. However, on October 29 we began to receive the 
replacement checks for the temporary rating.
    At this point in the timeline it is important to note that our 
family had now been without our full disability compensation and 
benefits for almost 11 months. Our college-aged children were forced to 
withdraw from school, and the overall financial strain, frustration 
level and emotional toll--in addition to the actual injury--were 
crushing.
    Finally, on January 7, 2008, after the intervention of Mr. Hall's 
Subcommittee and the Wounded Warrior Project, we received a final 
rating and back payment totaling thousands of dollars.
    As you can see we filed and re-filed, submitted and resubmitted, 
medical records, claims forms, applications, and so on, but no one 
seemed to be able to track anything, placing additional burdens on an 
already overwhelmed family. In our case, only after the intervention of 
a Congressional Office and a non-profit organization were we able to 
get the benefits Tai had earned. This process should not be this hard.
    Today, almost four years later, while we still have a few things to 
resolve with our rating and benefits, our family is trying to move on. 
Many people have stepped in to help from government agencies to 
Congressional offices to non-profit organizations. I am planning to 
return to work and school. Our children are returning to school, and 
Tai is enrolled in a media careers program for veterans in Chairman 
Filner's district. He has been a noted leader in the program and, ever 
the Gunny, has even spoken to the Wounded Warrior Project about being a 
peer mentor.
    However, our purpose in coming here is not only to tell you our 
story, but also to let you know that we are not alone. People we know 
have had similar problems, and we know there are more out there. We are 
hoping that our presence here will help you understand the obstacles 
faced by the wounded and their families and inspire all involved to 
work together to improve the efficiency of this vital system for the 
benefit of those who sacrificed so much for their country.
    Thank you and I look forward to answering any questions you may 
have.

                                 
     Prepared Statement of John Roberts, National Service Director,
                        Wounded Warrior Project
    Mr. Chairman, Ranking Member Lamborn, distinguished Members of the 
Committee, thank you for the opportunity to testify before you today 
regarding the use of technology to improve the efficiency of the 
Department of Veterans Affairs claims process. My name is John Roberts, 
and I am the National Service Director for the Wounded Warrior Project 
(WWP), a non-profit, non-partisan organization dedicated to assisting 
the men and women of the United States Armed Forces who have been 
injured during the current conflicts around the world. As a result of 
our direct, daily contact with these wounded warriors, we have a unique 
perspective on their needs and the obstacles they face as they attempt 
to transition and reintegrate into their communities.
    In addition to my experience with WWP in general and the 
Cleveland's specifically, I am a service-connected veteran, a former 
veterans service officer, and was most recently a supervisor with the 
Houston VA Regional Office where I reviewed claims and became familiar 
with a number of significant deficiencies within the system.
    In order to fully appreciate the problem, it is important to 
understand how the system currently operates. Despite recent advances 
in technology common to most businesses, the Veterans Benefits 
Administration (VBA) claims processing system is still dependent on a 
paper system. Although the VBA can now view electronic health records 
transmitted from the Veterans Health Administration (VHA), the ratings 
team is still required to print the records and place them in the 
veteran claims folder, which are then reviewed page by page by a Rating 
Veterans Service Representative.
    The current model of the VBA claims processing system has a total 
of six separate teams and often, but not always, includes another team 
that is dedicated to processing only the OIF/OEF cases. The six main 
teams are as follows:

      Triage which handles the incoming claims, evidence, and 
is charged with maintaining the outdated file cabinet system, which 
stores the hard copy paper claims files
      Pre-Determination which is charged with the initial 
development of all claims for service connected disability.
      Rating is responsible for reviewing all available 
evidence and determining if disabilities are service related. If so, 
the percentage of disability assigned.
      Post Determination is responsible for inputting awards 
and generating notification letters to the claimants.
      Appeals maintains all pending appeals submitted by the 
claimants.
      Public Contact is charged with the general phone calls 
and questions and conducting one-on-one interviews with the veterans, 
dependents and survivors.

    Files must be hand carried to each of the teams, and any member of 
these teams has access to the records at any given time.
    Despite the number of people with access and the ease with which a 
file may be misplaced, VBA has only one way to locate a claim file once 
it is removed from the filing cabinet. An electronic system called 
COVERS is available, but is only effective if utilized by the 
individual employee. Rather than having access to the file through 
electronic means, COVERS requires manual input to identify a specific 
location or individual. If this is not done, it is a very time 
consuming task to locate one file among all the files that are in the 
processing system. For example, in the Houston Regional Office (RO), 
there are approximately 200 employees and each person could have up to 
30 files or more on his/her desk.
    Another challenge is the outdated filing system used to store the 
thousands of active files warehoused either at or near each Regional 
Office. If a File Clerk or any employee for that matter is not paying 
attention and misfiles a claims folder into the wrong cabinet or 
drawer, it again becomes a very time consuming and difficult task to 
check each and every drawer to locate the missing file.
    The Triage team at each RO is also responsible for the intake of 
all new claims and all evidence submitted by each and every claimant. 
If the file is not easily located, it is placed on search within the 
COVERS system, until the file can be located. Because there are so many 
teams within the claims processing system, a particular file could be 
located within any of the teams at given time. This allows for the 
human error factor, which is often why the numerous pieces of vital 
evidence are often lost or misplaced and cannot be associated with the 
appropriate claim folder.
    If a file cannot be located and all avenues have been exhausted to 
locate the file, the RO will then take action to rebuild the folder 
from scratch. This means that all prior evidence and claims submitted 
by the claimant are also lost. The responsibility to replace the 
missing evidence or claims is placed on the claimant. The VA will ask 
the claimant to submit any copies that he/she may have in their 
possession.
    In addition, due to the current war on terrorism, VBA is faced with 
another challenge. The new challenge is trying to obtain records from 
National Guard and Reserve units. Active duty forces obviously do not 
file a claim until released from service. Once demobilized a member of 
the Reserve or National Guard component is eligible to file such a 
claim. If reactivated, however, the Reservist's claim is halted and he/
she takes the medical file with them to the theater.
    There is also the large backlog of records requests to the Records 
Management Center which houses not only claim folders, but now receives 
all service medical records for recently discharged servicemen. Think 
of this as a large warehouse of nothing but paper files and an 
inadequate staff to locate each and every file or record that has been 
requested by Regional Offices across the country.
    Another significant issue which can be identified at every Regional 
Office around the country is the varying levels of experience with the 
Rating Veterans Service Representatives (RVSR). In any given case, you 
could take five individual RVSRs and give them the same file and come 
up with five different opinions on how the case should be rated. 
Although there have been improvements with the implementation of RBA 
2000, the current electronic system utilized to rate compensation 
claims, the system is far from perfect. The overall ratings decision 
including service connection and actual percentage is left up to the 
interpretation of the individual RVSR. The gap in varying decisions 
nationwide can also be attributed to local policy at each individual 
Regional Office. While this has been the case for many years, the issue 
has come to a head due to the increased frequency with which this 
generation of veterans speak to each other and compare their individual 
situations.
    Mr. Chairman, unfortunately, these are only a few of the issues 
that surround a paper-based system, and situations like that of the 
Cleveland's are not unique. Many working groups, Government 
Accountability Office reports, and commissions have made 
recommendations on this topic. Most recently, the Veterans Disability 
Benefits Commission suggested that claims cycle times and accuracy 
could be improved by ``establishing a simplified and expedited process 
for well-documented claims, using best business practices and maximum 
feasible use of information technology.'' While the availability of 
well-trained, customer service minded employees cannot be overvalued, 
the implementation of recommendations such as these can help to greatly 
reduce the complexity of the claims processing system and result in 
timely results.
    WWP looks forward to working with you and the VA to try to resolve 
these problems. Thank you again for the opportunity to testify today, 
and I will be happy to answer any questions you may have.

                                 
 Prepared Statement of Tom M. Mitchell, Ph.D., E. Fredkin Professor and
    Chair, Machine Learning Department, School of Computer Science,
               Carnegie Mellon University, Pittsburgh, PA
Executive Summary
    Claims processing at the Veterans Benefits Administration appears 
to be amenable to a variety of improvements through the introduction of 
more computerized operations, including the adoption of artificial 
intelligence (AI) technologies for rule-based processing, case-based 
reasoning, and data mining.
    In the commercial sector, insurance claims are routinely processed 
online and automatically--Highmark, for example, processes over 90% of 
its insurance claims for hospital and physician services automatically, 
with no human intervention.
    To pick an example familiar to many, commercial software for filing 
income taxes (e.g., TurboTax) illustrates how computerization can 
improve accuracy, convenience and adherence to regulations when filling 
out complex forms and applying complex regulations automatically.
    The VA should be able to obtain similar benefits by computerizing 
its processing of claims. Much of this benefit can be achieved by 
applying well-understood computerized decision-support technologies 
that are already in widespread use in the commercial sector. Additional 
benefits may accrue from more advanced technologies that can be adopted 
once claims are captured and managed online.
    More specifically, three types of improvements to VA claims 
processing can be expected to follow from the adoption of online claims 
processing:

    1.  Shifting from pencil and paper claims to online claims can 
improve the accuracy, efficiency and convenience to veterans in filing 
and tracking their claims.
    2.  Introducing computer software to help interpret these online 
claims can improve the productivity of human claims processors, and the 
consistency and fairness of benefits awarded by (a) automating the more 
mundane and tedious steps in claims processing, and (b) informing human 
claims processors of benefits guidelines, typical awards, and similar 
past cases relevant to the claim they are evaluating.
    3.  Capturing the claims and their processing online provides 
additional opportunities for continuous and ongoing improvements to 
benefits processing, including (a) the use of data mining methods to 
predict and to flag new claims that are ``outliers'' likely to require 
collecting additional information, require specialized expertise for 
processing, etc., (b) the use of data mining methods to identify 
veterans who are not filing for benefits they should take advantage of 
given their condition, and to encourage them to seek these benefits, 
and (c) the use of historical claims and their disposition to help 
train newly hired claims processors.
Full Testimony
    Chairman Hall, and distinguished members of the committee:
    It is an honor to be asked to testify today, and to try to help you 
improve the situation faced by the brave men and women of our armed 
forces who have given so much on behalf of us all.
    My name is Tom M. Mitchell. I have been involved in the field of 
artificial intelligence for over 30 years, and am currently a Professor 
in the School of Computer Science at Carnegie Mellon University, and 
Chairman of the Machine Learning Department. While I can be considered 
an expert in computer science and artificial intelligence, I have no 
direct experience with the processing of veterans' benefits by the VA. 
Therefore, my comments should be interpreted as those of a technology 
expert who is not intimately familiar with the VA's claims processing 
system.
The Problem and the Opportunity
    That said, it is my understanding that we face a problem in the 
processing of benefits claims by the VA. I am told that the process of 
collecting and processing claims is primarily a manual, paper-based 
process in which 19 page claims forms are hand-written, and evaluated 
manually by claims processors. I am aware that there is an optional web 
interface at the Department of Veterans Affairs website for entering 
claims online, but that claims processing nevertheless remains 
primarily manual. I understand that in some cases delays of many months 
have occurred while veterans await the decision of the VA, and that 
recent increases in the number of claims have caused serious backlogs 
in processing, leading to undue hardships for our veterans.
    In my opinion, we have the technology we need to address and 
eliminate this problem.
    If we can develop computer software such as TurboTax, which guides 
taxpayers as they fill out complex tax forms online, and which then 
provides them with instant, computer based application of complex tax 
regulations to calculate to the penny the taxes they owe, then I see no 
reason why we cannot develop similar software to automate online filing 
of VA benefits claims and to automate a substantial fraction of the 
processing of these claims.
    To get even closer to the nature of the problem faced by the VA, 
consider current practices for processing medical claims in the 
insurance industry. The Senior Vice President for Health Plan 
Operations at Highmark Inc., which is a major provider of health care 
insurance in my own state of Pennsylvania, informs me that over 90% of 
the insurance claims they receive for hospital and physician services 
are processed fully by computer, with no human intervention.
    How do they do it? The insurance claim form is received 
electronically, and uses industry standard ICD9 codes to describe the 
hospital and physician services performed for the patient. The 
patient's eligibility is first checked automatically, by consulting a 
database describing which patients carry which benefits plans, and 
determining which policy was in force on the date the service was 
performed. The claim is then processed to determine the payment due, 
based on a combination of medical policy requirements, benefits already 
received by this patient, the specifics of their insurance plan, and 
the detailed information in the online claim form. In essence, the 
software is able to decide on the payment due because it contains a 
large collection of rules, each one specifying the payment to be made 
in some very specific case, defined by the details of the patient's 
policy, treatment, and history. The complex policy for determining what 
payment is due under which condition is encoded in these rules inside 
the computer. Once the software determines the payment due, the payment 
is issued automatically, and a record is added to the database to 
record this new bit of patient history--both the claim that was filed, 
and the outcome of the claim processing. Later, this database is used 
in a number of ways that further assist the patient and the insurance 
company. For example, by data mining the database, it is possible to 
detect that certain patients with chronic medical problems are not 
filing claims for medical treatments they should be receiving. In this 
case, Highmark can contact the patient to encourage them to seek this 
medical care. Highmark and other insurance companies have achieved 
great improvements in efficiency and effectiveness of claims processing 
by capturing the claims and the benefits policies online, and by 
automating the process of applying the policy to the claim.
    Can the VA do the same? While the type of benefits claims processed 
by the VA might not be identical to those processed at Highmark and 
other medical insurers, the two are sufficiently similar that one must 
conclude online processing will be of considerable value to the VA. A 
thorough analysis is needed to determine exactly which aspects of the 
VA claims processing are, in fact, amenable to computer automation, and 
which aspects require human judgment. I recommend that this analysis be 
performed immediately, and that a report be written that describes each 
step in the VA processing workflow, assessing the feasibility of 
automating or semi-automating each step. For those steps where human 
decision making is required, it may be the case that computer support 
for human claims processors could improve their effectiveness and 
efficiency, and this should also be examined in the report.
Relevant Computational and Artificial Intelligence Technologies
    Processing of online benefits claims may be improved by adopting a 
variety of well-tested computer technologies, including several 
artificial intelligence technologies that are routinely used today to 
support human decision makers. These technologies have been used, for 
example, to improve efficiency and effectiveness of medical claims 
processing in industry, to provide interactive support to workers in 
call centers who must gather information from customers and guide them 
to a solution, and to improve efficiency of filing and processing 
income taxes. All of these applications, like processing of VA 
benefits, involve collecting information from an individual and then 
making a decision based on a complex, predefined policy. Some of the 
key technologies used in practice in these and similar systems include:

    1.  Rule based decision aids. Rule-based software systems are 
widely used to encode expertise so that computers can apply this 
expertise automatically to assist in decision making. Rules are written 
in an ``IF-THEN'' format. For example, one fragment of the VA policy 
for evaluating claims can be captured in the following rule: ``IF there 
has been more than one episode of acute congestive heart failure in the 
past year, AND no chronic congestive heart failure THEN assign a 
disability rating of 60.'' Although described here in English, such 
rules are easily encoded in a computer language, then automatically 
applied to data on a benefits form. The VA benefits processing policy 
described in its ``Schedule for Rating Disabilities'' is described 
primarily as a very large collection of such rules.
    2.  Case-based reasoning systems. Case-based reasoning systems 
provide help to human decision makers such as call center personnel, by 
providing them rapidly with historical cases similar to the one they 
are currently processing, to help guide them as they process this new 
case. In the portion of benefits processing that requires human 
subjective judgments to evaluate the level of disability, it may well 
be helpful to the claims officer to examine the most similar past 
claims, as well as the judgments made in those cases. Case based 
reasoning is a technology that can quickly locate and deliver the 
relevant past cases from a database containing hundreds of thousands of 
historical cases, allowing it to act as an automated assistant to the 
human decision maker.
    3.  Machine learning and data mining. Machine learning algorithms 
and data mining systems that apply them to large databases are often 
able to discover important statistical regularities in the data that 
may not be apparent to a person. For example, large historical 
databases of credit card transactions are routinely mined to determine 
the features that indicate which future credit card transactions are 
likely to be fraudulent. In the VA claims database, data mining might 
be used to discover the pattern of features that indicate a claim will 
require additional information from the filer of the claim, or that a 
particular type of medical expertise will be required to evaluate it, 
or too, that the person filing the claim should also seek a particular 
additional preventative treatment. Data mining methods are widely used 
for applications where large numbers of historical records are 
available for computer analysis, from medical outcomes analysis, to 
telephone fraud detection, to targeted marketing to customers.
    4.  Methods for analyzing text and other unstructured data. 
Automated processing of information in forms is easiest if the fields 
in the form contain well-structured values (e.g., gender is either male 
or female, age is always a number). When a field in a form contains 
unstructured text (e.g., the description of how you were injured), 
automated processing is much more difficult because computers obviously 
cannot understand arbitrary sentences. Nevertheless, computer 
algorithms are able to extract some structured information from text, 
such as identifying names, dates, and locations that appear in the 
text. In some cases, this ability to extract certain types of 
structured data from text can transform the data into a form more 
amenable to automated processing.

    Technically, one can envision developing an approach for the VA in 
a sequence of stages. The first stage is to get claims online, so that 
they can be easily tracked and accessed. The second is to use well-
understood approaches to automating the easy steps in decision making 
(e.g., rule-based methods), and to provide interactive decision support 
to human claims processors when they must be involved (e.g., using case 
based methods to suggest relevant past cases for comparison and 
guidance). A third stage is to data mine the claims databases to 
analyze to improve claims processing, and to provide better service to 
veterans filing claims.
Conclusions and Recommendations
    In applications from insurance claims processing to tax filing to 
customer help centers, there is a growing and widespread use of 
computer-based tools to capture forms online, and to automate some or 
all aspects of processing these forms. The VA should take advantage of 
this well-developed computer technology, to move its own benefits 
processing online, in order to gain a number of likely benefits 
including:

    1.  Shifting from pencil and paper claims to online claims can 
improve the accuracy, efficiency and convenience to veterans as they 
file and track the processing of their claims.
    2.  Introducing computer software to help interpret these online 
claims can improve the productivity of human claims processors, and the 
consistency and fairness of benefits awarded by (a) automating the more 
mundane and tedious steps in claims processing, and (b) presenting 
claims processors with the past claims that are most similar to the one 
they are currently processing, to help guide their analysis.
    3.  Capturing the claims and their processing online provides 
additional opportunities for continuous and ongoing improvements to 
benefits processing, including (a) the use of data mining methods to 
predict and to flag new claims that are ``outliers'' likely to require 
collecting additional information, require specialized expertise for 
processing, etc., (b) the use of data mining methods to identify 
veterans who are not filing for benefits they should take advantage of 
given their condition, and to encourage them to seek these benefits, 
and (c) the use of historical claims and their disposition to help 
train newly hired claims processors.

    Given the urgency of addressing current backlogs in processing VA 
benefits claims, the VA should mount an aggressive effort to move its 
benefits processing online. Three specific steps can be performed in 
parallel to move rapidly toward this goal:

    1.  Begin immediately a process to move all VA benefits claims 
forms into an online database, and to maintain the status of claims 
processing and its final disposition. This will be useful for managing 
and tracking claims even before processing of claims is automated.
    2.  Conduct a detailed study of the workflow process used by the VA 
to evaluate benefits claims, producing a report that describes each 
step in the workflow, and whether and how this step can be automated. 
For workflow steps that require human subjective judgments, the study 
should describe how computer decision aids can provide support to the 
human.
    3.  Consult with large insurance companies and others who process 
benefits claims to understand current best practices and to begin a 
process of adopting them where appropriate.

    Mr. Chairman, addressing the problem of effective and efficient 
processing of benefits claims does not require new technological 
breakthroughs--it requires the adoption of current best practices that 
are based on well-understood technologies. I hope my remarks are useful 
to your deliberations, and I thank you for the opportunity to present 
my views here today.
References
    1.  TurboTax software for automatic completion of tax forms. 
Available free online at http://turbotax.intuit.com/
    2.  Department of Veterans Affairs, Book C Schedule for Rating 
Disabilities, available online at http://www.warms.vba.va.gov/
bookc.html#f

                                 
    Prepared Statement of Randolph A. Miller, M.D., Donald A.B. and
    Mary M. Lindberg University Professor of Biomedical Informatics,
    Medicine, and Nursing, Vanderbilt University School of Medicine,
                             Nashville, TN
Overview
    1.  My comments describe the applicability of Biomedical 
Informatics to improving the processes for determining Veterans' 
eligibility for disability compensation.
    2.  Clinical informatics involves information management and 
decision making during healthcare delivery. Expert systems, which 
utilize artificial intelligence techniques, represent a subset of the 
more general decision support techniques and electronic health record 
approaches that clinical informatics provides.
    3.  I generally endorse the findings of the June, 2007 Institute of 
Medicine Report, ``A 21st Century System for Evaluating Veterans for 
Disability Benefits''. That report lacks adequate detail in some areas 
pertinent to clinical informatics. I clarify below how informatics can 
make a difference.
    4.  Clinical informatics can improve both the speed and quality of 
disability determinations for U.S. Veterans. The highly acclaimed VistA 
and CAPRI systems developed by the Veterans Health Administration (VHA) 
and Veterans Benefits Administration (VBA) provide excellent examples 
of relevant clinical informatics applications. Future informatics 
efforts can help in five important areas by:

      a.  converting paper-dependent processes to electronic processes 
for all stages of disability determination and designating VistA as the 
definitive repository for such information, with supporting software 
tools such as CAPRI available in whatever venues are appropriate;
      b.  creating electronic definitions for each of the approximately 
700 disability conditions, based on current CFR 38 definitions and 
amendments, that utilize standardized terminology for concepts and 
findings (e.g., the SNOMED-CT, LOINC, DSM, and ICD terminologies);
      c.  creating a tracking system that captures electronically all 
information relevant to each Veteran's disability determination. The 
system would provide VBA raters with a checklist to determine what 
documents they require, a method to order necessary tests and 
procedures, and a dashboard to indicate the status of each document 
needed to complete a review. These would help raters to locate and 
retrieve the information efficiently, as well as to determine what 
information was incomplete or missing, and limit unnecessary duplicate/
repeated testing that delays disability determinations and increases 
costs;
      d.  creating decision support tools to identify and display 
disability-specific patient information to VBA raters in a manner that 
allows them to determine easily which disability criteria have been met 
or not met, and to recommend appropriate next actions electronically; 
and,
      e.  creating a quality feedback loop using current information to 
evolve future practices through ongoing continuous improvement.

    5.  In matters as important as Veterans' disability determination, 
computer-based tools, including ``expert systems'', can serve as 
adjuncts to help humans to collect and manage information, but the 
tools cannot in any way replace the most important aspects of human 
judgment. Present and future computer-based tools will not displace the 
talented, experienced people who comprise the present VHA and VBA. 
Informatics can help people to work ``smarter'', in order to benefit 
Veterans.
    6.  A key lesson from clinical informatics is that while change can 
be beneficial, it can also be disruptive if employees must dramatically 
and abruptly alter their work processes. Whatever plan of action the 
government adopts to move from current state to a more ideal future, 
the plan must be pragmatic, and coordinated to proceed in concrete, 
non-disruptive steps. Each step must convey benefits and lay the 
foundation for subsequent steps with greater benefits. Changes must be 
gradual and familiar to already overburdened employees. This document 
describes such scenarios.

    Figure 1 illustrates applicability of the above ideas and 
principles to various stages of disability determination: definition of 
disability, documenting conditions during active military duty, 
documentation of health and disability data within the Veterans Health 
Administration system; and, determination of disability status by VBA 
raters.

[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]


    I will describe the applicability of biomedical informatics to the 
task of improving the processes and workflows used to determine 
Veterans' eligibility for disability compensation. Clinical informatics 
involves information management and decisionmaking during healthcare 
delivery. Expert systems, which utilize artificial intelligence 
techniques, are a subset of the more general decision support 
techniques and electronic health record approaches that clinical 
informatics provides.
    I would first like to place the role of computer-based decision 
support tools in proper perspective. As I detailed in 1990 in an 
article in the Journal of Medicine and Philosophy,1 for 
matters as important as Veterans' disability determination, computer-
based tools, including ``expert systems'', can serve as adjuncts to 
help humans to collect and manage information, but the tools cannot in 
any way replace the most important aspects of human judgment. If a 
patient grimaces and winces while struggling to walk across a room, and 
then claims, ``I'm OK, doc'', a compassionate human can correctly 
categorize the patient's condition, but an expert system is far less 
likely to do so if it only has ``I'm OK, doc'' as the patient 
description. Present and future computer-based tools will not displace 
the talented, experienced people who comprise the present VHA system. 
Nevertheless, the VHA system can achieve significant progress and 
efficiency through greater application of electronic information 
systems in determining and tracking Veterans' disability benefits. I 
generally endorse the findings of the June, 2007 Institute of Medicine 
Report, ``A 21st Century System for Evaluating Veterans for Disability 
Benefits''. That report lacks adequate detail in some areas pertinent 
to clinical informatics. I clarify below how informatics can make a 
difference. Clinical informatics can improve both the speed and quality 
of disability determinations for U.S. Veterans. The highly acclaimed 
VistA and CAPRI systems developed by the VHA provide excellent examples 
of relevant clinical informatics applications. I will refer to the 
principles and recommendations listed in Figure 1 and explain each.
PHASE ONE--BUILD INFRASTRUCTURE TO AUTOMATE STEPS OF EXISTING 
        DISABILITY DETERMINATION SYSTEM, TO GAIN EFFICIENCY
    The first task is to identify disability-related records and 
documents that exist now in paper format and to convert them, whenever 
possible, into a standardized, electronically processable format. As 
noted in Figure 1, Recommendation 1.2, this involves creating 
standardized names for all document types used in disability 
determination (there is already a history of being able to do so within 
certain segments of the VHA system--see, for example).2 A 
review of a sufficient number of existing disability records (paper and 
electronic) used to establish (or deny) each disability category for 
hundreds to thousands of Veterans would help to create a complete list. 
Having specific names for each type of document makes it easier for VBA 
raters to find and manipulate them.
    In changing to a more electronic disability determination system, 
one must be careful to convert almost all routine activities of VHA 
raters to being electronically based, with actions analogous to what 
they now do with paper. It would potentially be worse--more cumbersome 
and slower for disability determinations--if VBA raters had to use both 
paper and electronic record systems in processing a Veteran's 
application, than to use only one or the other system. If both paper 
and electronic systems were in active use, a VBA rater would always 
have to check both systems to see if ``missing'' items in one system 
are actually not ``missing'', but present in the second system.
    The most straightforward way to begin conversion to electronic 
processing is to identify where paper records are currently generated, 
and where existing VHA software is applicable to creation of electronic 
versions of that paper-based information. For all other paper records 
that cannot be easily converted in this manner, the document naming 
system should be used to label them, and then they should be 
electronically scanned to create electronically retrievable records. 
Such record types should be scheduled for subsequent projects to 
capture them at their source--at time of generation--using future 
electronic capture tools analogous to CAPRI. The award-winning CAPRI 
system developed by the VHA provides templates that prompt physicians 
and other clinicians as they examine a patient for disability 
determination, and stores the information in a standard form within the 
VHA's VistA electronic medical record system. Recommendations 2.1, 2.2, 
3.1, and 4.1 suggest that the DoD, the VHA healthcare providers, and 
the VHA disability raters rapidly move toward 100% utilization of the 
current version of CAPRI to capture disability-related information as 
it is generated, in all situations where CAPRI is applicable.
    The next stage of utility for automated systems to enhance VBA 
raters' processing of disability claims would be, per Recommendation 
1.1, to categorize the criteria required to establish each of the 
approximately 700 disability conditions specified in CFR 38 and its 
amendments.
    To develop electronic criteria for each disability condition would 
require human review of the latest version of CFR 38 and amendments for 
each condition, and creating: (a) a list of findings, coded in SNOMED-
CT or LOINC, required to be present to establish the disability, (b) a 
list of findings, coded in SNOMED-CT or LOINC required to be absent to 
establish the disability, (c) a list of findings that help to support 
the presence of the condition but which are not required to establish 
the condition, coded in SNOMED-CT or LOINC, (d) a list of the document 
types (using the standardized document names per Recommendation 1.2) 
that are relevant to determination of the specific disability 
condition, (e) the list of CAPRI frame identifiers that are relevant to 
determination of this specific disability, (f) names for each of the 
700 conditions coded wherever possible in ICD, DSM, or SNOMED-CT, and 
(g) narrative text that describes the remaining criteria for the 
establishment of the specific disability condition that could not be 
coded in steps (a) through (c).
    Once the electronic identifiers exist for information relevant to 
disability determination, including the CFR 38 definitions, the 
document names, the finding names, and the CAPRI template IDs, it is 
possible to create an electronic tracking system, per Recommendation 
4.2, that can indicate for VBA raters which documents and findings are 
required to establish the disability, and what the status of each is 
for a given Veteran applying for disability. An electronic dashboard 
that displays the status and availability of each item of information 
could then be constructed. It is possible that aspects of the dashboard 
might be shared with the Veterans who apply for disability through the 
``my HealtheVet'' web portal created by the VHA.
    An adjunct to the above system would expand CAPRI to assist VBA 
raters in ordering the best tests and procedures to complete disability 
determination efficiently, and to record the reasoning and conclusions 
the VBA raters used to establish or deny the specific disability claim.
    The recent Institute of Medicine report, ``A 21st Century System 
for Evaluating Veterans for Disability Benefits'' (National Academies 
of Science Press, 2007; Copyright  National Academy of Sciences. 
http://www.nap.edu/catalog/11885.html) recommends: ``Recommendation 6-
1. VA and the Department of Defense should conduct a comprehensive 
multidisciplinary medical, psychosocial, and vocational evaluation of 
each veteran applying for disability compensation at the time of 
service separation.'' This should initially be done using the CAPRI 
system in its current state, recording the results in a VistA 
compatible format for future reference at VHA and VBA.
    The above-described steps are somewhat straightforward, and are 
within the reach of existing technology, although they require 
substantial effort in terms of system development, security and 
confidentiality assurance, application testing, training of end users, 
and ongoing technical support. The steps essentially comprise a basic 
level of automation of the current disability determination process in 
a manner that will assist VBA raters in carrying out their work more 
efficiently. Once such an infrastructure is established, substantial 
enhancements could be made, some of which would involve simple decision 
support techniques, and others of which would involve machine learning 
and expert system approaches.
PHASE TWO--ENHANCE AUTOMATED INFRASTRUCTURE FOR DISABILITY 
        DETERMINATION WITH DECISION SUPPORT FEATURES
    A number of techniques developed over the past three decades for 
clinical decision support 3-10 are relevant to future 
enhancements to a VHA/VBA disability determination and documentation 
system. At the national level, the VHA has been a major contributor to 
clinical decision support through its evolution of the VistA electronic 
medical record system. In addition, many talented individuals working 
within the VHA and VBA have also made contributions.
    One important technological approach is clinical diagnostic 
decision support systems,3-6 which can be probabilistic 
(Bayesian), criterion-based, or heuristic (``artificial intelligence'' 
expert systems) 3 in nature. In general, such systems take 
as input standardized vocabulary descriptors describing a patient's 
condition (such as history, physical examination, or laboratory 
findings) and produce as output a ranked list of possible diagnoses and 
a suggested approach to determining which diagnoses are present.
    A second important ``expert system'' technique relevant to clinical 
informatics is natural language text processing.8-10 Using a 
target vocabulary of defined clinical terms or concepts, such as 
provided by the U.S. National Library of Medicine's Unified Medical 
Language System Metathesaurus, or by the SNOMED-CT terminology system 
officially endorsed by the U.S. Government, such programs can scan a 
``free text'' document, such as a clinical note, and identify which of 
the target concepts are present in the document.9 The 
utility of such an approach for VBA disability determination has 
already been demonstrated by a pilot project to identify spinal injury 
related findings from free text disability exam records, and to 
correlate those findings with an electronic representation of the 
criteria used by VBA to determine disability.10
    Finally, ad hoc or heuristic approaches can combine manual 
techniques with semi-automated approaches to characterize clinical 
domains or conditions.11-12 Such approaches have been used 
to derive a standardized vocabulary for patients' problem lists from a 
large set of examples in free text,11 and to attempt to 
convert information stored in disparate DoD and VHA clinical record 
systems from one representation format to the other.12
    The recent Institute of Medicine report, ``A 21st century System 
for Evaluating Veterans for Disability Benefits'' (National Academies 
of Science Press, 2007; Copyright  National Academy of Sciences. 
http://www.nap.edu/catalog/11885.html) contained the following 
recommendations:

          ``Recommendation 3-1. The purpose of the current veterans 
        disability compensation program as stated in statute currently 
        is to compensate for average impairment in earning capacity, 
        that is, work disability. This is an unduly restrictive 
        rationale for the program and is inconsistent with current 
        models of disability. The veterans disability compensation 
        program should compensate for three consequences of service-
        connected injuries and diseases: work disability, loss of 
        ability to engage in usual life activities other than work, and 
        loss in quality of life.''
          ``Recommendation 4-1. VA should immediately update the 
        current Rating Schedule, beginning with those body systems that 
        have gone the longest without a comprehensive update, and 
        devise a system for keeping it up to date. VA should 
        reestablish a disability advisory Committee to advise on 
        changes in the Rating Schedule.''
          ``Recommendation 4-6. VA should determine the feasibility of 
        compensating for loss of quality of life by developing a tool 
        for measuring quality of life validly and reliably in the 
        veteran population, conducting research on the extent to which 
        the Rating Schedule already accounts for loss in quality of 
        life, and if it does not, developing a procedure for evaluating 
        and rating loss of quality of life of veterans with 
        disabilities.''

    The effort to redefine the conditions for which disability 
compensation is appropriate should be standards-based (ICD, DSM, 
SNOMED-CT, LOINC) as described above. Text-mining and natural language 
processing methods could be used to determine which coded terms are 
currently used in disability determinations through review of the 
thousands of existing electronic disabilty-related VistA and CAPRI 
records, and from samples of paper records converted by OCR or direct 
typing into electronic format. This review, coupled with the effort to 
extend disability criteria as recommended by the IOM Report, could 
result in computer-processable ``criteria table'' definitions for each 
disability condition that would maximize the objective representations 
of each condition (while still retaining free text if necessary to 
describe the aspects of human judgment required in each determination). 
As previously recommended, the list of document types and procedures 
relevant to determination of each disability category, as well as the 
orders required to carry out the procedures in VistA, could be added to 
an expanded revision of CAPRI.
    Once the above representation scheme for each disability condition 
was in place, an expert system using the ``criteria table'' approach 
could be developed to assist VBA raters in determining the completion 
status of each disability determination, and added to a more advanced 
version of the previously mentioned dashboard system. The AI-RHEUM 
expert diagnostic system,6 developed in part at the U.S. 
National Library of Medicine, might be used as a starting point for the 
proposed VHA/VBA expert system.
    A similar system could be developed for use within the DoD 
electronic medical record system, which would employ natural language 
processing and expert criteria table methods to identify portions of an 
active duty service individual's record that would suggest eligibility 
for disability evaluations before discharge from active duty, per the 
IOM Report recommendation, ``Recommendation 6-1. VA and the Department 
of Defense should conduct a comprehensive multidisciplinary medical, 
psychosocial, and vocational evaluation of each veteran applying for 
disability compensation at the time of service separation.''
PHASE THREE--CREATE A QUALITY FEEDBACK PROCESS TO ENHANCE AND EVOLVE 
        THE DISABILITY RATING PROCESS OVER TIME
    The recent Institute of Medicine report, ``A 21st Century System 
for Evaluating Veterans for Disability Benefits'' (National Academies 
of Science Press; Copyright  National Academy of Sciences. http://
www.nap.edu/catalog/11885.html) contained the following 
recommendations:

          ``Recommendation 4-2. VA should regularly conduct research on 
        the ability of the Rating Schedule to predict actual loss in 
        earnings. The accuracy of the Rating Schedule to predict such 
        losses should be evaluated using the criteria of horizontal and 
        vertical equity.''
          ``Recommendation 4-3. VA should conduct research to determine 
        if inclusion of factors in addition to medical impairment, such 
        as age, education, and work experience, improves the ability of 
        the Rating Schedule to predict actual losses in earnings.''
          ``Recommendation 4-4. VA should regularly use the results 
        from research on the ability of the Rating Schedule to predict 
        actual losses in earnings to revise the rating system, either 
        by changing the rating criteria in the Rating Schedule or by 
        adjusting the amounts of compensation associated with each 
        rating degree.''
          ``Recommendation 5-1. VA should develop a process for 
        periodic updating of the disability examination worksheets. 
        This process should be part of, or closely linked to, the 
        process recommended above for updating and revising the 
        Schedule for Rating Disabilities. There should be input from 
        the disability advisory committee recommended above (see 
        Recommendation 4-1).''
          ``Recommendation 5-3. VA should establish a recurring 
        assessment of the substantive quality and consistency, or 
        inter-rater reliability, of examinations performed with the 
        templates and, if the assessment ends problems, take steps to 
        improve quality and consistency, for example, by revising the 
        templates, changing the training, or adjusting the performance 
        standards for examiners.''

    Once a fully electronic system was available that both represented 
criteria for disability determination electronically, and which 
recorded individual Veteran's records in standardized terminologies, 
text mining and machine learning techniques could be used to accomplish 
the above-mentioned IOM objectives, and to provide feedback for 
quality-based evolution of the proposed systems.
BACKGROUND INFORMATION
    My own background is that I trained and became Board-certified in 
Internal Medicine during the 1970s. For over a quarter century, I cared 
for inpatients and outpatients in academic settings, including in 
several VHA Hospitals and Clinics. My research over the past three 
decades has been in the area of clinical informatics. I was the 
founding Chief of the Section of Medical Informatics at the University 
of Pittsburgh, and the founding Chair of the Department of Biomedical 
Informatics at Vanderbilt. For more than two decades, I helped to 
develop and evaluate expert systems for medical diagnosis at the 
University of Pittsburgh. After moving to Vanderbilt in 1994, I helped 
to develop clinical decision support tools implemented within 
Vanderbilt's homegrown care provider order entry (CPOE) system. That 
CPOE system improves quality of care, safety, and cost-effectiveness by 
giving advice to physicians in real-time as they care for patients.
    During my career, I have been Principal Investigator on over $20 
million of Federal grants and contracts related to biomedical 
informatics. I am currently a member of the Institute of Medicine of 
the National Academies of Science. I am a Past President of the 
American Medical Informatics Association (AMIA) and a Past President of 
the American College of Medical Informatics (ACMI). I currently serve 
as Editor-in-Chief of the Journal of the American Medical Informatics 
Association (JAMIA). Two of the clinical informatics systems that I 
helped to design and build have been disseminated commercially. My 
comments above have no direct relationship to those commercialization 
efforts, and I have no conflicts of interest in that regard.
REFERENCES
     1.  Miller RA. Why the Standard View is Standard: People, not 
Machines, Understand Patients' Problems. J Med Philosophy. 1990; 
15:581591.
     2.  Brown SH, Lincoln M, Hardenbrook S, Petukhova ON, Rosenbloom 
ST, Carpenter P, Elkin P. Derivation and Evaluation of a Document-
naming Nomenclature. J Am Med Inform Assoc. 2001;8(4):379-390.
     3.  Duda RO, Shortliffe EH. Expert Systems Research. Science. 1983 
Apr 15;220(4594):261-268.
     4.  Miller RA, Pople HE Jr, Myers JD. INTERNIST1, An Experimental 
Computerbased Diagnostic Consultant for General Internal Medicine. N 
Engl J Med. 1982; 307:46876.
     5.  Bankowitz RA, McNeil MA, Challinor SM, Parker RC, Kapoor WN, 
Miller RA. A computer-assisted medical diagnostic consultation service: 
implementation and prospective evaluation of a prototype. Ann Intern 
Med. 1989; 110:82432.
     6.  Miller RA. Medical diagnostic decision support systems--past, 
present, and future: a threaded bibliography and brief commentary. J Am 
Med Inform Assoc. 1994;1(1):8-27.
     7.  Aliferis CF, Miller RA. On the heuristic nature of medical 
decision-support systems. Meth Inform Med. 1995 Mar;34(1-2):5-14.
     8.  Tom Mitchell. Machine Learning, McGraw-Hill: New York, NY. 
1997.
     9.  Uzuner O, Goldstein I, Luo Y, Kohane I. Patient Smoking Status 
from Medical Discharge Records. J Am Med Inform Assoc. 2008; 15: 14-24.
    10.  Brown SH, Speroff T, Fielstein EM, Bauer BA, Wahner-Roedler 
DL, Greevy R, Elkin PL. eQuality: Electronic quality assessment from 
narrative clinical reports. Mayo Clin Proc. 2006; 81(11):1472-1481.
    11.  Brown S, Miller RA, Camp H, Giuse D, Walker H. Empirical 
Derivation of an Electronic Clinically Useful Problem Statement System. 
Ann Intern Med. 1999; 131(2):117-126.
    12.  Bouhaddou O, Warnekar P, Parrish F, Do N, Mandel J, Kilbourne 
J, Lincoln MJ. Exchange of Computable Patient Data Between the 
Department of Veterans Affairs (VA) and the Department of Defense 
(DoD): Terminology Standards Strategy. J Am Med Inform Assoc. e-
published Dec 20, 2007 doi:10.1197/jamia.M2498

                                 
  Prepared Statement of Marjie Shahani, M.D., Senior Vice President, 
           Operations, QTC Management, Inc., Diamond Bar, CA
    Mr. Chairman, Members of the Subcommittee, thank you for the 
opportunity to testify before you today on this important topic of 
claims processing. QTC is a nationwide, private provider of medical 
examination and medical record services to the medical and disability 
communities, including Federal, state and local government agencies; 
property and casualty insurance carriers; third-party administrators; 
employers--and the claimants they serve. With 580 highly trained 
employees located at five strategically placed regional administrative 
offices, 35 owned and operated medical facilities in seven states, and 
more than 12,000 pre-screened medical professionals, QTC has produced 
more than 2.5 million quality medical exams and reports over the past 
ten years and pioneered software and technology to ensure quality, 
timely, customer-focused and cost effective services for our clients 
throughout our 25 years of experience.
QTC Experience with VBA C&P Process
    QTC has been a provider of C&P examinations to the Veterans 
Benefits Administration (VBA) since 1998. QTC provides medical 
examinations to veterans and active duty servicemembers seeking 
compensation from the VBA in 12 VA Regional Offices (VAROs) and 40 
Department of Defense (DoD) Benefits Discharge Determination (BDD) 
sites around the country. After completing the medical exam, QTC 
provides a detailed narrative report according to the guidelines of the 
VA's Automated Medical Information Exchange (AMIE) worksheets. VA's 
rating specialists then use this information along with the veteran's 
claim file, or c-file, to adjudicate the veteran's claims. Supporting 
the VBA for the past nine years has provided QTC with an in-depth 
understanding of the complexity of the activities, nuances and 
uniqueness of the VA medical disability examination process, as well as 
its challenges.
QTC Developed Software Applications to Support the C&P Process
    QTC has applied its experience and expertise in developing 
knowledge-based tools and technology, not as the company's primary 
purpose, but to facilitate and streamline the work of QTC's medical 
professionals and support staff.
    Specifically for veterans' C&P medical examinations, QTC has 
developed:

      A Client Portal--an Internet-based application 
specifically designed and written for VA Central Office, VAROs, and 
DoD. This application provides VA and DoD with online access to the 
status of each case and appointment details along with the exam 
results, including the narrative report and diagnostic test results, 
and provides the ability to review the progress of any case, review 
management reports, case summaries and perform a batch report download 
from a secured Internet connection.
      A Protocol Engine--A medical exam protocol engine, or 
software, that synthesizes multiple VA required AMIE worksheets with 
portions of the VA rating schedule (38 CFR, Part 4) into one veteran 
specific medical examination form, thereby providing examiners with a 
single head-to-toe examination. It organizes the VA requirements into a 
form that assures the physician will address each and every medically 
claimed condition for each veteran. This software generates a specific 
exam protocol based on VA AMIE worksheets and the veteran's claimed 
conditions, and it allows for immediate updating to conform to VA 
statutory, regulatory and medical changes and quick dissemination to 
all providers nationwide.
A Logical Next Step--Organize the Evidence
    Secondary to our nine years of experience working with VA, we 
acknowledge the complex challenges of the VA medical disability 
process--it is like no other disability program we have worked with 
previously. In an attempt to provide value-added services, QTC applied 
its knowledge and experience to the next step in the C&P process--
specifically simplifying and streamlining the information gathering 
process for the rating specialists so that they are able to rate the 
veteran's claim in a quality, timely, customer-focused and cost-
effective manner.
    QTC developed an Evidence Organizer (patent-pending)--a working 
prototype that is an automated rating tool designed to assist the VA 
rating specialist and significantly reduce the time to determine a 
rating decision. The Evidence Organizer has great potential to help 
rating specialists search and find relevant information needed to 
determine and rate claims and thus help the VBA facilitate the process 
of adjudicating each claim.
Current Rating Process Productivity and Challenges
    The rating process begins when a veteran files a claim. The VA 
Triage Team identifies the type of disability claim. Then the VA Pre-
Determination Team determines the need for additional information, such 
as a medical examination, and gathers the additional records. Once all 
the evidence is gathered, the Rating Team reviews the entire contents 
of the c-file and rates the veteran's claim, determining entitlement to 
benefits and the degree of disability. The Evidence Organizer was 
designed to assist the rating specialist in their process of reviewing 
the entire contents of the c-file and to assist them in making that 
final decision.
    Currently, the VA goal for each rating specialist is to rate an 
average of 3-4 cases per day. In 2006, QTC interviewed experienced 
claims examiners and asked them to break down the process and allocate 
time for each step in the rating process, assuming seven claimed 
conditions. The four steps of the rating process and allotted times 
were:

    1.  Initial review of the c-file: 20 minutes.
    2.  Linking the evidence in the file to the claimed conditions: 85 
minutes.
    3.  Determining the severity of the condition and if each is 
service connected: 40 minutes.
    4.  Writing the rating decision: 65 minutes.

    The total time to examine a veteran's claim is 210 minutes or 3.5 
hours. It becomes evident that a rating specialist cannot meet a target 
of 3-4 cases per day in one 8-hour workday. In fact, the actual average 
number of cases a rating specialist can process in one day is 2-3 
cases.
    In 2007, based on QTC's data, the average number of conditions 
claimed per veteran was four for C&P cases and eight for BDD cases. 
Examples of multiple-claimed conditions include left knee pain, asthma, 
low back strain, foot fungus, hearing loss, and depression. 
Additionally, reopened claims now account for 54% of all open cases, 
and cases are being reopened with additional unclaimed secondary 
conditions, such as hypertension, headaches, or scars.\1\
---------------------------------------------------------------------------
    \1\ Statement of Mr. Ronald R. Aument before the House Committee on 
Veterans' Affairs, March 13, 2007.
---------------------------------------------------------------------------
    The biggest challenge the rating specialist faces is finding the 
medical evidence in the veteran's paper files supporting the claimed 
conditions and linking the evidence to the appropriate rating code as 
listed in the 38 CFR, Part 4. QTC's Evidence Organizer would eliminate 
this challenge, allowing the rating specialist to more efficiently make 
the rating determinations. The Evidence Organizer accomplishes this by 
creating an organized electronic c-file, providing the rating 
specialist the ability to quickly search and review all available 
evidence at the click of a mouse. The rating specialist no longer needs 
to review the entire c-file over and over again for each claim they are 
rating as the Evidence Organizer has organized all the available 
evidence for them.

[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]

Potential Impact
    The Evidence Organizer decreases the decision time to rate veterans 
claims from 3.5 hours to 2.2 hours, a time savings of 37% per decision, 
increasing the number of veterans' cases rated from 2 per day to 3 per 
day. On an annual basis this would increase the number of claims 
decisions per rating specialist to 711 from the current 533, an 
increase of 178 decisions per rating specialist or a 33% increase.
Evidence Organizer Process
    The Evidence Organizer can be applied to all four steps in the 
rating process, decreasing time spent per case file by organizing and 
highlighting all medically related information. The Evidence Organizer 
works by converting the cumbersome paper-based claim file (c-file) to 
create an electronic record or file (e-file). This document management 
process begins with a Technician scanning in the c-file and other hand 
written documents through the use of Optical Character Recognition. The 
software transforms each record into a text searchable digital record. 
As additional records become available they are also integrated into 
the e-file. At the heart of this process is QTC's core knowledge 
database, which is built upon our extensive disability examination 
experience supporting the VBA C&P examinations.
    The knowledge database identifies, highlights and electronically 
indexes all keywords and claimed conditions, for example: diabetes, 
asthma, arthritis, as well as any potential claimable conditions 
throughout each record, thereby providing the rating specialist with 
all possible claimable conditions. Once the e-file has been 
established, each record is reviewed to validate the software's 
indexing, creating an initial table of contents for the e-file.
    The next step involves a Reviewer validating the highlighted 
records and linking the referenced medical evidence to the VA rating 
requirements in 38 CFR, Part 4. PDF scanned records not compatible with 
electronic screening methods (handwritten records) are reviewed page-
by-page by the Reviewer and relevant information is highlighted, 
extracted, and digitally indexed and linked to the rating criteria 
appropriate for the claimed condition or potentially claimable 
condition. Once all the records have been reviewed the software creates 
a full and complete e-file with a table of contents listing all claimed 
conditions.
    Finally, the complete annotated e-file is electronically available 
for the VA rating specialist to review and assist in their rating 
decision process. The software suite allows the VA rating specialist 
to:

      Review and search each and every document at the click of 
a mouse.
      Review all tagged, annotated and associated data.
      Add the rating specialist's determination of relevance 
with rationale electronically.
      Identify, tag and index additional information as 
desired.
      Document the rating decision made with the referenced 
evidence.
      Review any additional potentially claimable conditions

Summary
    By applying new technologies such as the Evidence Organizer, the 
VBA could:

      Organize medical evidence.
      Reduce routine and repetitive tasks.
      Increase accuracy by facilitating cross-referencing.
      Link the rating criteria with the clinical annotation.
      Scan, index and review all medical records.

    Also, The Evidence Organizer will potentially complement RBA-200 or 
other existing VBA software and serve as a training tool. With the 
current challenges the VBA is experiencing in claims processing, the 
application of this technology is essential to improving performance 
and efficiently providing veterans with accurate ratings.

                                 
                  Prepared Statement of Ned M. Hunter,
     President and Chief Executive Officer, Stratizon Corporation,
                  Atlanta, GA, (VA State Pilot Study)
    Chairman Hall, Ranking Member Lamborn, and distinguished Members of 
the Committee, thank you for the opportunity to appear before you 
today.
    Stratizon Corporation is a veteran owned Software-as-a-Service 
company, which has utilized the concepts of artificial intelligence to 
successfully design a software platform and application solely focused 
on improving the VA's claims processing system.
    We have gained valuable insight into the underlying success of 
using AI to solve the VA's claims processing system. First, the 
technology available in the marketplace is adaptable, flexible, 
scalable, proven, and cost-effective. Technology is not to be resisted 
but embraced. Second, success will be highly dependent upon the 
perspective in which AI solutions are constructed. A true veteran-
centric solution must be constructed through the eyes and situation of 
the veteran to satisfy the requirements of the state and Federal 
policies and VA systems and not constructed through the eyes of the 
multiple government entities to independently present the bureaucracy 
to the veteran.
    Stratizon applied this perspective in successfully piloting for the 
U.S. Navy, three unique web-based ``intelligent'' solutions that 
demonstrated how the quality of life for sailors could be significantly 
improved by replacing confusing, complicated, paper intensive, and 
manually driven enterprise processes with web-based, easy-to-use, 
automated, and complete self-service solutions, or what we define as 
``intelligent user interfaces'' or ``IUIs''. ``IUIs'' can also be 
designed for numerous veteran events such as transitions from active to 
veteran status or applications and appeals for VA compensation and 
health benefits.
    The Commonwealth of Virginia's Department of Veteran Services, 
working with the Joint Leadership Council of Virginia representing 32 
veteran service organizations, is implementing such a solution called 
TurboVetTM. Building on a successful pilot in 2007, the 
Governor of Virginia has included funds in his FY 09 budget that begins 
on July 1, 2008, for full production. TurboVetTM will 
provide Virginia veterans, or an authorized representative, the ability 
to log online at Virginia.gov, via a personal computer or device such 
as this Apple IPhone, and select an event they need assistance with. 
Initially a series of statements and questions regarding their status 
or particular event will be presented. Their personal data currently on 
file with the state will be retrieved so they may confirm or validate 
their data, thus improving data integrity and eliminating redundant 
data entry. The system will use embedded decision logic to react 
intelligently to their input to continually refresh and display only 
the necessary event questions, thus eliminating the frustration of 
redundant and unnecessary questions. A list will be displayed of all 
state and Federal benefits the veteran has earned with all 
corresponding documents spanning multiple agencies required for the 
veteran to submit, thus providing a peace of mind to the veteran their 
solution is holistic. Each document will then be progressively, 
simultaneously, and perfectly auto-populated with the proper data, thus 
eliminating data transcriptions errors and numerous processing delays. 
Finally, the veteran will have the option to save and print each 
document locally and, at their discretion, electronically submit their 
data securely to all participating authorities and systems to be 
processed and tracked fully and completely.
    Virginia's success in using an AI platform is dependent upon the 
continued support and cooperation of all parties, both political and 
technical. Decisionmakers need to remain committed to this paradigm 
shift to the future and must always provide the best institutional 
knowledge available to ensure the TurboVetTM 
IUITM not only becomes that benchmark of service but also 
remains the benchmark. We need technical cooperation between state 
agencies to take advantage of TurboVetTM's ability to 
seamlessly exchange data with disparate IT systems. We need cooperation 
and support at the Federal level. Federal supervisors in Roanoke have 
projected that a minimum of 100 days of processing time will be 
eliminated when the TurboVet system is implemented at only the state 
level. Stratizon foresees few problems in exchanging data between 
TurboVetTM and VA systems such as VistA and VetsNet. We 
fervently believe there could be significant process cycle time 
improvement and extraordinary cost savings at the state and Federal 
level if veteran's data at the state level could first be ``pre-
verified'' against recognized ``authoritative'' national VA databases 
and then seamlessly exchanged upon claims submission and during the 
claims management process. Virginia's goal is to fulfill the vision of 
H.R. 3047 and have a claim prepared properly with attached medical 
evidence and documentation for electronic submission to Federal 
adjudicators for rating, and have those claims calculated fairly, 
consistently, and automatically.
    In summary, using a properly designed AI system would dramatically 
improve the VA claims processing systems by improving the access to 
customer solution and service for veterans and their family members, 
reducing the costs to the state in staff administration, training, and 
paperwork, and improving the accuracy, throughput, and expediency of 
claim submissions by the state for VA adjudication.
    On behalf of the Stratizon Corporation, I would like to thank the 
Chairman and all committee members for this opportunity to be here 
today.

                               __________
    In today's world the gathering, administration and management of 
employee data has become a serious issue for corporate America. In 
particular, when an employee is confronted with a business or personal 
life event, the employee is required to provide sensitive personal data 
to his or her employer or to corroborate the data the employer already 
has on file for the employee. The employer, on the other hand, is under 
great pressure and exposure to liability to process the event and the 
data surrounding it swiftly, correctly, confidentially and completely. 
In most cases, the individual providing the data and the organization 
processing the data waste an extraordinary amount of time and energy 
ensuring all personal data associated with the event is properly, 
accurately and securely processed.
    For the individual, incorrect data processing can mean endless 
hours of wasted time, frustration, and even financial loss. For the 
organization, incorrect data processing means significant and 
unnecessary overhead expense due to decreased business productivity and 
additional time and materials to correct data errors and to account for 
lost data. Ultimately, the corporate liability for improperly or poorly 
managing the business or personal life event for the employee is an 
unacceptable business risk which can harm the employee/employer 
relationship and damage the reputation of the organization.
    The solution for this challenge is for enterprises to migrate from 
enterprise employer centric solutions to customer employee centric 
solutions through the use of open-standards web-based XML technology 
and artificial intelligence concepts. Stratizon Corporation has engaged 
this vision and has developed an open-standards J2EE technology 
platform that creates software applications to display on-screen an 
Intelligent User Interface (IUITM).
    An Intelligent User Interface (IUITM) is an online 
series of questions, consolidated from all the questions from a variety 
and multitude of forms and applications pertaining to a specific 
business or life event, and embedded with decision logic encapsulating 
the major life event, to manage the event for the user by intelligently 
responding to that user's input. In conjunction with accessing and 
utilizing the user's data stored in the enterprise in multiple and 
disparate systems, a complete picture materializes for the use and a 
total solution is rendered.
    The process begins when a user logs onto a website or access device 
such as an Apple IPhone and selects a button that describes the 
personal event they are confronted with. The IUITM firsts 
presents the data the organization has on file for the user and allows 
the user to confirm or change the information. The IUITM 
then presents a series of questions regarding the event to gather the 
required missing data the organization needs from the user so the 
organization can process the event for the user and the organization
    This allows the IUITM to quickly complete all steps of 
the event process, auto-populate all required documents with the option 
to print or digitally sign, and send the data to the enterprise in the 
proper format to process the event. The result is a life event 
processed simply, correctly and completely. The benefits are the 
reduction in human frustration, elimination of paper, event processing 
time and administrative data management costs for both the user and the 
organization.
    These customized intelligent user interfaces become the ``edge of 
the enterprise,'' presenting a less bureaucratic image, providing a 
less bureaucratic process to match, and providing several advantages.
    First, there is a remarked improvement in end user satisfaction and 
customer service because solutions have been engineered with a focus on 
intelligence event solutions and not just intelligent document 
solutions. It is more about the data than the document. This innovative 
approach to user-centric design significantly reduces user frustration 
and costs by more efficiently and cost effectively integrating, 
delivering and managing the entire process and the requisite data, 
forms and documents required from end users.
    Second, an AI platform delivers better collective ``time to value'' 
for end users and organizations because it is sophisticated yet simple 
in its presentation. By designing a solution from an end user's 
``holistic event'' perspective versus an enterprise's ``silo-driven'' 
perspective, we ensure an event is managed with 100% completeness for 
the user and the enterprise.
    Third, an open standards XML platform has better flexibility in 
working with existing corporate systems. Today's data systems need to 
integrate with multiple web servers, application servers and database 
server architectures and seamlessly integrate into an organization's 
portal strategy. For the enterprise, this means lower cost, easier 
scalability, and faster development and delivery of a solution.
    In summary, an open standards XML artificial intelligent platform 
provides a powerful development environment for rapidly developing and 
deploying electronically enabled intelligent user interface 
applications to provide the following benefits:
Consumer/Family Benefits
      Improves end user productivity and morale
      Reduces the pain associated with life-event related 
paperwork
      Eliminates difficult searches across multiple Web sites 
and departments
      Reduces need for forms
      Ensures the proper data is completed and submitted 
accurately and timely
      Improves end user satisfaction with organization service 
and support
Organization Benefits
      Streamlines data and forms processing via the web
      Establishes a consistent interface for all forms-based 
events
      Reduces administrative costs of providing service
      Reduces training costs of customer support and service 
personnel
      Reduces personnel time to manage forms inventory and life 
event packets
      Reduces redundant data entry by customer support and 
service personnel
      Improves data integrity for the organization information 
technology systems
      Reduces process cycle-times
      Reduces process re-work through increased data integrity
      Provides workflow capabilities
      Reduces cost of maintaining and handling supply of paper 
forms

                                 
    Prepared Statement of John F. McGarry, Senior Vice President of 
                               Benefits,
                 Chief Risk Officer, Unum, Portland, ME
    Mr. Chairman, members of the Committee, I'd like to thank you for 
the opportunity to testify before you today. My name is Jack McGarry 
and I am the Senior Vice President of Benefits and Chief Risk Officer 
at Unum.
    I am here today to discuss how our technology facilitates claim 
management (case management) decisions at Unum.
    We process approximately 400,000 disability claims per year and pay 
about $4 billion in benefits directly to our insureds and their 
families. Most of Unum's claims are governed by ERISA, the Federal law 
which generally requires insurance companies to make disability claim 
decisions within 45 days. Unum's experience shows that it is possible 
to manage high volumes of claims in a timely and accurate manner while 
achieving high levels of customer satisfaction. In fact, 93% of our 
customers report that they are satisfied with the overall quality of 
contact with Unum. Ninety-six percent are satisfied with our timeliness 
and ability to respond to their questions. And 97% find us courteous 
and respectful.
    Technology is an important component of the solution to managing 
the volumes and timeframes, as well as our customer service. However, 
the decision about a person's ability to work is also informed by in-
depth analysis of pertinent documents and discussions with claimants, 
their employers and their physicians in order to assess their ability 
and motivation to work.
    In the end, the disability determination is a judgment call that 
needs to be made by people.
    In order to assure that the right people are reviewing the right 
claims at the right time, a combination of Unum's technology and people 
is necessary. For example, a routine claim may be automatically sent by 
the system to one person while a complex claim with multiple diagnoses 
may go to another based on a combination of systems and management 
decisionmaking. As robust as our systems are, a person does look at 
every claim we pay.
    Our technology, operated by our people, does the following:

      Manages documents
      Facilitates workflow
      Ensures a complete administrative record, and
      Monitors and measures quality and service results

    1.  First, our system manages documents. Our files can grow to 
hundreds--if not thousands--of pages. With our image-based system: All 
files are paperless, multiple people can access the same claim at the 
same time, and documents are organized and stored in an efficient 
manner. This reduces redundancies in workload management. For example a 
nurse, a claim payer and a vocational rehabilitation specialist can all 
be working on the same file at the same time--even if these people are 
located in different parts of the country.
         Another efficiency of the system is that our paperless files 
can be viewed all at once or electronically ``tabbed'' in different 
ways--for example, all the medical data can be viewed using a medical 
view of the file. This creates efficiency in that if a doctor needs to 
review a file, she is able to view the relevant medical data, and does 
not have to review data unrelated to the medical condition unless 
necessary. The file can be viewed using multiple other tabs--for 
example, the letters can be viewed, financial data, or vocational 
information. Thus, the technology frees our people to view only the 
data pertaining to the issue they are working on--allowing each claim 
payer to focus on critical case management activities.
    2.  Second, our system facilitates workflow. All information is 
electronically scanned into our system upon receipt. The act of 
scanning documents as they are received creates an online activity for 
the claim payer to review. In our system, every action a person 
completes creates another action or follow-up activity. The system can 
also trigger an action for someone to review claims and/or contact 
customers at key times during the claim management process.
         This technology allows for activities and discussion to be 
focused on ongoing claim management. With real time access to 
information, our team focuses on the document review and discussion 
needed to facilitate next steps on the claim instead of recapping 
redundant data or gathering additional information that may not add 
value to the process. Real time access involves multiple users being 
able to access and update the claim file at the same time (parallel 
claim processing). It also means that as documents are received via 
mail or fax and scanned into the system, they are immediately viewable 
by all involved in the claim. For example, a doctor may be writing a 
report at the same time a claim payer is checking whether the claimant 
is contractually eligible for benefits--the moment the doctor completes 
her report, the claim payer would be able to see it in the image based 
system and begin working on the medical case as well.
         The image based claim management system also contains a 
letter-writing function and a letter library so that a claim payer can 
write to claimants and their doctors. Addresses populate automatically, 
saving the claim payer time. Also, the claim payer can set follow-up 
activities so the claims stay on track. An example of a follow-up 
activity may be that the claim payer sets an automatic reminder for the 
system to let the claim payer know whether requested data has arrived 
after a certain number of days.
    3.  Third, our system ensures a complete administrative record. An 
administrative record is important for ERISA purposes as well as sound 
claim management. When a claim changes hands between claim payers, all 
of the management activities associated with that claim--including 
future activities--stay with the claim and are automatically assigned 
to the new claim payer. The technology keeps the file together in one 
place and minimizes any disruption in service due to a personnel 
change. Thus, if a file does change hands, many redundant steps in 
getting a new person up to speed on the file and setting direction are 
eliminated.
    4.  Fourth, our system enables us to monitor and measure quality 
and service results. Management and our Quality Assurance process 
require the ability to review files ``real time''--at the same time 
that the claim payer is working on the files. The system automatically 
tracks and reports our service times and outcomes.
         Because each one of the activities the claim payer does is 
scheduled and tracked, we can ensure that the right resources are 
applied to the right claims at the right time. Thus, our management and 
Quality Assurance teams can provide feedback as the claims are being 
managed--as opposed to days, weeks or months later.

    We separate levels of disability into those which have shorter 
durations--what we call Short Term Disability (STD--6 months or less in 
duration)--and those which may be longer term--what we call Long Term 
Disability (LTD--may last greater than 6 months)--based on a number of 
factors--including experience of the claim payer and diagnosis.
    At the initial level, for the shorter terms claims: Our in-take 
department reviews each new claim and assigns an ICD-9 (International 
Code of Diagnostics 9) diagnosis code. We use the ICD-9 system because 
it is a standardized system within the medical community. Using a 
standardized system avoids confusion and makes it easy and efficient to 
communicate with the claimant's medical team about the condition and 
treatment recommendations.
    Thus, when a claim is filed, we use a quick assessment--a triage-
type process--to assess the case based on the employee's diagnosis and 
other relevant criteria. Our operating standard for short term claims 
is that at least 95% of our claims have been paid, denied or pended--
with the status and reason communicated to the claimant--within 5 days 
of our receipt of a complete claim.
    Long term claims are more complex and as a result require more 
extensive investigation and skilled resources. Most of these claims 
take longer than 5 days to assess. Thus, the system assigns claims to a 
manager based on certain criteria and the manager gives different 
numbers of claims to people based on data elements including diagnosis 
and the expertise of the claim payer.
    For all claims, we may gather information from multiple sources 
including the claimant, the employer, the claimant's medical team and 
our own medical and vocational resources. There is continuous 
communication throughout this process.
    After the initial assignment, our technology initiates reports 
based on key measures--including diagnosis, generally accepted medical 
duration guidelines, and our Unum database information. These reports 
can identify claims that need additional work or follow-up, and help 
each claim payer to determine what steps to take next.
    Let's take a look at how a hypothetical claim may work through the 
process. As news reports have widely documented that many veterans 
suffer from behavioral health issues, I will use an example involving a 
claimant with depression.
    In the behavioral health context, our experience shows it is 
critical for the well-being of the insured that the benefit specialist/
case manager be proactive and get involved in helping the claimant 
return to functionality quickly. Speed, accuracy and quality are 
critical, not only for the health and well-being of the claimant, but, 
as previously mentioned, because most of our claims are governed by 
ERISA, which among other things requires private sector insurance 
companies to make disability claim decisions within 45 days. While 
there are some provisions which do grant additional time under specific 
circumstances, generally speaking we must make our decisions within 45 
days.
    On the complex behavioral health claim example, on Day One our 
Intake Department would set up the claim on the image based system and 
scan the documents into the electronic file.
    On Day Two the system assigns the claim to a claim payer based on 
the ICD-9 code and other criteria.
    On days three through five, having access to the file, the claim 
payer reviews the specific facts of the claim, including checking 
contractual eligibility and determining whether a decision can be made. 
A decision to pay could be made, for example, if the person is 
hospitalized in an in-patient setting.
    If more data is needed, the claim is pended and the claimant is 
kept informed of the status. A claim may be pended if relevant medical 
information has not been received yet. When that information is 
received, a decision is reached by the claim payer or by the claim 
payer working with the appropriate clinician or resources through what 
we call our ``roundtable process,'' which would include physicians, 
managers, and vocational rehabilitation consultants.
    In our example, after payment has begun, our system is designed to 
proactively identify claims for further attention based on a number of 
triggers. If the system identifies these triggers--for example it may 
identify a claimant with depression that is either approaching or 
passing the timeframe in which recovery would be expected--it schedules 
a claim management activity so the right person reviews the file on the 
date specified and can initiate or delegate the appropriate activity. 
The triggers may be identified based on this claimant's diagnosis and 
our own duration management database. The activity may be for a Unum 
nurse to call and work with the claimant, for example, ensuring the 
claimant is getting the appropriate treatment needed for recovery.
    If the claimant's symptoms are still acute at the six week mark, we 
would continue to pay and work with the claimant and the insured's 
medical team. On an ongoing basis we would stay in regular contact with 
the insured and make sure we understand the continued nature and 
severity of the condition.
    At around the 60-day mark, the claim would be transitioned to a 
different claim payer with more in-depth expertise and training in 
managing complex claims. This claim payer would again keep in regular 
contact with the insured, and would continue to work closely with 
medical and vocational resources. If the claimant were hospitalized or 
had suffered a severe injury--the claim may be sent to an extended 
duration unit, where follow up would be less frequent.
    During this phase the following steps could occur:

      continued evaluation of the claimant's functional 
ability, which could include setting up an independent medical 
examination;
      in depth assessment of the physical and cognitive 
occupational demands of the insured's occupation or other occupations 
they may be suited to perform;
      vocational assistance, determining any possible 
accommodations that could be made so the person can return to work; and
      continued partnership with in-house medical, vocational 
and management resources as needed.

    In summary, the critical data elements associated with specific 
claims are identified in short and long term situations and our benefit 
specialists and/or nurses are able to focus on the right activity at 
the critical time.
    Ultimately, technology can facilitate claims handling processes and 
decisionmaking by helping ensure that the documents are being reviewed 
by the right people at the right time. It can also provide cost 
savings. We estimate that since we implemented this system we have 
saved 10-20 per cent over the previous system. More importantly, it has 
increased claim management effectiveness and allowed us to pay claims 
more accurately.
    In the end, however, it is the skill of the management and the 
people handling the claims, supported by those working with them, who 
are responsible for the claim and service to our customers. Through the 
time savings and other efficiencies created by technology, each claim 
handler is freed to devote more time to interact directly with 
claimants and their physicians in building a plan and assessing a 
person's ability to return to work.
    Disabilities present a complex management challenge because they 
are logistically difficult, judgment based and can be emotionally 
charged. Technology can help facilitate judgment based decisionmaking 
but we don't see it as ever being able to replace people in the claim 
management process.
    I would like to end by extending an invitation to all of you and 
for VA staff to visit Unum and would welcome the opportunity to 
continue to be a resource for sharing best practices between the public 
and private sectors as you continue to evaluate the disability 
adjudication/case management process. Thank you for the opportunity to 
testify before the Committee.


[GRAPHIC(S) NOT AVAILABLE IN TIFF FORMAT]

   Prepared Statement of Gary A. Christopherson, University Park, MD
   (Former Senior Advisor to the Under Secretary for Health and Chief
Information Officer, Veterans Health Administration, U.S. Department of
 Veterans Affairs, and Former Principal Deputy Assistant Secretary for
              Health Affairs, U.S. Department of Defense)
    Chairman Hall and Mr. Lamborn, let me applaud you for holding these 
hearings on substantially improving the claims processing systems and 
greatly improving how we care for our Nation's veterans. Today's 
hearing, ``The Use of Artificial Intelligence to Improve the VA's 
Claims Processing System'' is very important in its own right but more 
so in how it supports the overall effort toward a truly caring, timely 
and effective processing of our veterans' disability claims.
    In my testimony today, I will be speaking both to the true meaning 
and obligation of ``the duty to assist'' and to the strong enabling 
role that artificial intelligence and other decision support tools can 
and should play. I will point to the fact that, in the military where 
our veterans served, great honor is given to those who deliver ``on 
time and on target''.
    Why do I believe this is so important? I saw the sacrifice of our 
servicemembers and the incurred debt by our Nation when I had the honor 
of serving as the Principal Deputy Assistant Secretary of Defense for 
Health Affairs. I saw and helped achieve the great potential that the 
Military Health System has for truly caring for our servicemembers, our 
retirees and their families.
    Subsequently, I saw the plight of many of our veterans and the 
great obligation of our Nation when I had the honor of serving as the 
Veterans Health Administration Chief Information Officer and Senior 
Advisor to the Under Secretary for Health. I saw and helped achieve the 
great potential that the Veterans Health System has for truly caring 
for our veterans. In this latter role, I also had the opportunity to 
work with the Veterans Benefits Administration as they explored the 
potential for a much better claims processing system.
    Most importantly, I had the great privilege of getting to know both 
servicemembers and veterans as people providing a great service to our 
Nation. I had the opportunity to work closely with organizations 
supporting active servicemembers, the Guard, the Reserve, retirees, 
veterans and their families.
    All this taught me that what the Department of Veterans Affairs can 
and should be doing is about the veteran and that everything VA does 
should be centered around the veteran. VA should be a truly veteran-
centric system, including its claims processing system. This is not 
about claims per se, it is about a veteran who needs the support of the 
Nation that he or she served.
    This is about strong leadership. This is about effective management 
end-to-end. This is about deploying effective technology. Most 
importantly, this is about a new process that honorably discharges more 
of our obligation and delivers benefits on time and on target to our 
veterans.
A New Claims Processing System
    Let me start with the claims processing system. If we believe that 
veterans are hurting, that we have ``the duty to assist'', and that we 
should meet the expectation of being ``on time and on target'', then we 
need a new process now. My suggestions for one such proposal and its 
rationale are detailed in the footnote at the end of my statement.\1\
    Several years ago, I had the opportunity to help advise the 
veterans benefits system in its thinking about a new system. I asked 
them how many hours of work it takes to process a claim. They said 
about eight hours on average. Sadly, this means it takes six months to 
a year or more to complete eight hours actual work. That makes no 
sense.
    What does make sense is a new system operating in real time like we 
do with the health care system. When a veteran is hurting and needs 
health care, the veterans' health system assists the veteran and 
provides care quickly. When a veteran is hurting and needs financial 
benefits, the veterans' benefits system does little to assist, forces 
the veteran to navigate a large bureaucracy and massive paperwork, and 
provides financial benefits only after months or years.
    VA staff should come out and welcome the veteran. They should 
actively assist the veteran to get everything processed quickly and 
correctly. They should be working with the veteran on an ongoing basis 
as case managers.
    But we also need the continuing support and assistance of the 
Veterans Service Organizations. They have played an invaluable role 
over the years in trying to support the veteran in a broken system.
    Changing the process means giving a veteran the financial benefit 
at least as soon as the veteran files a claim with basic evidence 
supporting that claim. Real time would mean that VA would receive the 
claim and supporting evidence and make the decision on the same day. 
That is achievable if we really want to make it happen.
    Let's change the assumptions. Let's have VA presume that a veteran 
filing a claim with basic evidence supporting that claim is entitled to 
the associated benefit a) for the period of one year or b) until VA 
completes the processing of that claim [no more than one year], 
whichever is shorter. Further, let's have VA begin payments to a 
veteran within 30 days of that veteran filing a claim with the 
supporting evidence.
    But let's also have the VA claims processing system function 
better. Within six months, let's have VA institute a new claims 
processing system that proactively assists a veteran with his/her 
claim. Let's have them produce a temporary or permanent decision 
(preferred) within two weeks of a veteran filing a claim with basic 
evidence supporting that claim. And, let's begin payments to a veteran 
within 30 days of that veteran filing a claim with the supporting 
evidence. Further, preferably within six months, let's have VA deliver 
the permanent decision within two weeks.
    This can be done. It should be done.
Enabling Technology
    In my colleagues' testimony, we have heard that the technology 
exists today to greatly improve the speed and accuracy of benefits 
decisions. Using artificial intelligence or electronic decision support 
tools is nothing new. Both government and the private sector use them 
every day.
    For those who argue benefit claims processing is a much more 
complicated and difficult process, I counter that it is not. If you 
want to go to the most complicated and difficult process that exists 
today, it is arguably health care. Yet, somehow health care has figured 
out how to provide care in real time without technology and even better 
with technology.
    One doesn't even have to go outside the VA system to find real time 
systems. The veterans' health care system is just that. Much praise has 
been given to the veterans' health system for its responsiveness, 
quality and effective use of technology. Today, the veterans' health 
care system can access any electronic information in any VA health care 
site in real time. With due diligence, this will be even better in the 
future. One of my honors was to have both rescued the VistA health 
information system and to have set it on the path to an even brighter 
future.
    To help improve the claims processing system, we even went one step 
further. We made the veterans health care information available to the 
claims processing system electronically and in real time. 
Unfortunately, it doesn't seemed to have sped up claims processing.
    Artificial intelligence and human intelligence together can make a 
difference. Using health care as an example, we have some of the most 
highly trained people in the world providing health care. More and more 
we are using less highly trained people as well. More and more, these 
talented people rely upon many kinds of decision support tools to help 
them be effective. They have real time access to virtually everything 
there is to know about a person that can help improve health. More and 
more, they have real time access to the most current knowledge on any 
particular condition and on caring for the whole person. They are 
supported by guidelines and advice on what best will help a person get 
healthier.
    Sometimes they treat even without full information if that is what 
it takes to reduce the misery now. Sometimes they have to gather 
information from other physicians across town or across the country. 
Sometimes, more often now than in the past, they have all the 
information they need to treat within their own health information 
system. VA's VistA and the future HealtheVet VistA do just that. Many 
public and private sector health care providers' health information 
systems do just that.
    These health care providers have the decision support necessary to 
care for a person in real time. They don't have to wait months or years 
to make decisions on how to treat the person in misery today. The 
person doesn't have to wait months or years to get help with their 
misery.
    But, keep in mind that health care did not wait for technology; it 
just got better with technology as an enabler. Technology can greatly 
enable the claims processing system, but it is only an enabler. For 
better claims processing, we don't have to wait for the technology. We 
can start reducing the misery today and then do it even better when the 
technology arrives.
Leadership and Management
    However, getting to a new, veteran-centric, effective claims 
processing system with the necessary enabling technology will only 
happen if VA leadership is fully committed. This is not going to be 
easy, but it is doable.
    VA leadership will have to adopt the vision of a system that gets 
needed and appropriate support to the veteran in real time, meaning 
days and not months or years. They need to lead a claims processing 
system that welcomes the veteran at the door (preferably reaches out to 
them earlier) and treats the veteran as one who has earned that 
welcome. They need to lead a system where the intent is to have the 
veteran file a claim and get approval (if appropriate) on the same day. 
They need to lead the current staff on how to better use their skills 
to help the veteran. They need to demonstrate leadership and work in 
partnership with the VSOs on how to make the systems better and on how 
the VSOs can continue to assist the veteran during claims processing 
and afterward.
    But leadership will not succeed without effective management to 
make and continue to make all this happen. Not just VA but most of the 
Federal Government does not have the management strength. It clearly 
doesn't have enough to deal with the level of change we are suggesting 
here. For VA, there is good management talent in the veterans benefits 
area. VA will need to develop that talent as well as bolster that 
talent with strong managers from outside the veterans' benefits system 
as well.
Duty to Assist; Expectation to Be on Time and on Target
    As I conclude, I am sure that some will argue it is unaffordable or 
undoable. Let me suggest not. First, it could well be built into the 
$100-$150 billion economic stimulus package moving at this very moment. 
Second, let me remind you that when we send our service members to war 
we seem to handle the budget. We can and should do no less when they 
come home and need our help. That is part of the real cost of 
preventing or conducting war.
    Today, there is a failure to understand and appreciate the 
veteran's plight. Today's claims processing behavior is more like a 
castle under siege rather than a home providing compassion, warmth, 
help, and sustenance. Contrast that with the veterans' health system 
where care is provided in real time with most administrative details 
sorted out later. Feel what it is like for a veteran to live in 
uncertainty and without support for six months or a year or more. What 
if we did that for health care? It would be unacceptable. Protests 
would ring in and outside of every care facility. Why do we tolerate it 
for benefits determination?
    For a better future, the bottom line is this. Change the 
assumptions. Change the process. Use the best technology. Care for the 
veteran. Meet our obligation--the duty to assist. Deliver on time and 
on target. ``The duty to assist'' is an obligation that VA, with regard 
to benefits, has yet to honorably discharge. ``On time and on target'' 
is what we expected of our veterans and what we should expect of VA.

                               __________

         \1\ Economic Stimulus and Duty to Assist Our Veterans

What economic stimulus?
    Effective upon enactment, the Department of Veterans Affairs shall:

      Proactively assists a veteran with his/her claim,
      Presume that a veteran filing a claim with basic evidence 
supporting that claim is entitled to the associated benefit a) for the 
period of one year or b) until VA completes the processing of that 
claim [no more than one year], whichever is shorter, and
      Begin payments to a veteran within 30 days of that 
veteran filing a claim with the supporting evidence.

    Within six months after enactment, the Department of Veterans 
Affairs shall institute a new claims processing system that:

      Proactively assist a veteran with his/her claim,
      Produces a temporary or permanent decision (preferred) 
within two weeks of a veteran filing a claim with basic evidence 
supporting that claim, and
      Begins payments to a veteran within 30 days of that 
veteran filing a claim with the supporting evidence.
Why this economic stimulus?
      Stimulus is temporary (budget impact), timely and on 
target.
      Dollars paid to veterans will go directly and quickly 
into the economy to cover basic living expense.
      Housing foreclosures will be reduced as veterans are able 
to stay current on their mortgages.
      The misery of our veterans with disabilities and awaiting 
a benefits decision will be substantially reduced as the backlog is 
essentially eliminated.
      The Department of Veterans Affairs will be incentivized 
to move as soon as possible to a ``real time'' claims processing 
system.
      An important symbolic and real step would be taken to 
address the needs of veterans with current and future disabilities 
returning from Iraq and other wars.
      An important symbolic and real step would be taken to 
keep the promise to care for our servicemembers when their service 
results in disabling conditions. This also has a positive impact on 
recruitment and retention.
      The Nation would finally be really moving toward meeting 
the obligation of ``Duty to Assist''.
      Bipartisan support is highly likely.

                                 
   Prepared Statement of Kim A. Graves, Director, Office of Business 
                                Process
   Integration, Veterans Benefits Administration, U.S. Department of
                            Veterans Affairs
    Mr. Chairman and Members of the Subcommittee, it is a privilege to 
be here today to talk about the use of information technology to 
enhance claims processing within the Veterans Benefits Administration 
(VBA).
    VBA has made significant strides in the use of information 
technology to improve claims processing in all of our benefit programs. 
Our current focus is the development of a comprehensive strategy to 
integrate the various initiatives already underway, leveraging 
successes already accomplished. VBA is collaborating with the Office of 
Information and Technology (OI&T) in developing this strategy to ensure 
our mission needs are met and that the appropriate enterprise 
architecture is employed.
    At the core of our strategy is the implementation of a business 
model for Compensation and Pension processing that is less reliant on 
paper documents. The use of imaging technology and computable data to 
support claims processing in our Insurance, Education and Loan Guaranty 
programs has been successful for many years. Initial pilot efforts in 
our Compensation and Pension business line have demonstrated the 
feasibility of using this type of technology for these benefit programs 
as well.
    Our comprehensive strategy, the Paperless Delivery of Veterans 
Benefits initiative, is envisioned to employ a variety of enhanced 
technologies to support end-to-end claims processing. In addition to 
imaging and computable data, we will also incorporate enhanced 
electronic workflow capabilities, enterprise content and correspondence 
management services, and integration with our modernized payment 
system, VETSNET. In addition, we are also exploring the utility of 
business rules engine software for both workflow management and to 
potentially support improved decisionmaking by claims processing 
personnel. A recent Request for Information yielded a variety of 
products that may be useful in our end-state vision.
    As part of our strategy for improving the claims processing 
business model, VBA recently contracted with IBM to conduct a study of 
the current process and suggest improvements. We expect their report 
shortly and will assess their findings as we move forward with 
documenting our strategy.
    As noted previously, two pilot programs are currently underway and 
have demonstrated the utility of imaging technology in our Compensation 
and Pension business line. Both projects utilize our Virtual VA imaging 
platform and related applications. Virtual VA is a document and 
electronic claims folder repository.
    The first pilot supports our income-based pension program. It 
involves imaging documents received in conjunction with the annual 
income verification and reporting process. This imaging allows the 
three Pension Maintenance Centers (PMCs) to make the necessary claims 
adjustments without need for retrieval and review of the paper claims 
file.
    The second pilot supports the compensation program at the 
centralized rating activity sites for our Benefits Delivery at 
Discharge (BDD) program. The separating servicemember's medical records 
and supporting claim information are imaged at the outset of the claims 
process. This allows rating veterans service representatives to make 
decisions based solely upon review of the imaged records without 
recourse to a paper claims file. Further refinements of the business 
process are now underway to identify gaps in the existing system 
capabilities which will enhance our understanding as we evaluate 
options for expanding use of this technology.
    An additional pilot project is also under development. This project 
will examine issues such as user authentication and using online forms 
to provide the capability for the initial ``electronic'' filing of 
benefit claims. This is the first step in implementing online ``self-
service'' to allow veterans to manage some of their interactions with 
VA electronically.
    Integration with VETSNET is also a critical success factor in our 
overall strategy. We have made significant progress in the 
implementation of VETSNET over the past two years. Approximately 98 
percent of all original compensation claims are being processed end-to-
end in VETSNET, and we are now paying monthly compensation benefits to 
more than 850,000 veterans--or approximately one out of every three 
compensation recipients--using this modernized platform. With our next 
conversion of records from the legacy Benefits Delivery Network (BDN), 
scheduled for April, VETSNET will be the primary payment system for 
Compensation and Pension benefits.
    Integration and data exchange with the Department of Defense are 
also essential, as is our continued expansion of exchange of health 
care information with the Veterans Health Administration.
    In 2001, the Compensation and Pension Records Interchange (CAPRI) 
application was developed jointly by VHA and VBA. CAPRI provides VBA 
claims processing personnel access to information from the Veterans 
Health Information Systems and Technology Architecture (VistA) and that 
is used in the development and documentation of disability benefit 
claims. CAPRI also provides access to some Department of Defense 
medical records through integration with the Federal Health Information 
Exchange (FHIE) framework. As part of our vision for the future, we 
will eventually move this data directly into our paperless benefits 
delivery platform as part of the veteran's ``e-file.''
    As we continue to move forward with the efforts described here, we 
are focused on developing an integrated project plan, ensuring the 
needs of our veterans and their families are documented and attainable. 
Demonstrable milestones and performance metrics will be incorporated so 
that we and our stakeholders are able to assess our progress in 
achieving our vision.
    To assist in developing this plan, we are working closely with our 
OI&T partners to develop a Request for Proposals to engage the services 
of a Lead Systems Integration contractor. The integrator will provide 
support in documenting both the business and technical requirements for 
implementation of our strategy.
    I assure you the Under Secretary for Benefits is committed to 
implementation of the Paperless Delivery of Veterans Benefits 
initiative. Together with our partners in the Office of Information and 
Technology, we believe this goal is not only attainable, but is 
imperative to ensure the best possible service to our Nation's 
veterans. Thank you for the opportunity to address these important 
issues.
    Mr. Chairman, this concludes my prepared testimony. I would be 
happy to answer any questions you may have.

                                 
  Prepared Statement of Stephen W. Warren, Principal Deputy Assistant
  Secretary for Information and Technology, Office of Information and
            Technology, U.S. Department of Veterans Affairs
    Mr. Chairman and Members of the Subcommittee, I would like to thank 
you for the opportunity to testify today, on the use of information 
technology to enhance claims processing, within the Department of 
Veterans Affairs (VA), as well as utilizing data from the Veterans 
Health Information Systems and Technology Architecture (VistA), to 
assist in the processing of disability claims. These are very important 
issues that affect the life of every veteran and their just 
compensation for disabling injuries, received while serving our 
Country.
    I would like to begin by addressing VA's efforts at leveraging 
information technology to improve the timely delivery of veterans' 
benefits. The Office of Information and Technology (OI&T) has been 
collaborating with the Veterans Benefits Administration (VBA), in the 
development of a comprehensive strategy to enable the achievement of 
their target business model. The operational concept of the Paperless 
Delivery of Veterans' Benefits initiative is to employ enhanced 
technology platforms to include imaging, computable data, electronic 
workflow capabilities, and enterprise content and correspondence 
management services. The initiative will integrate with VBA's core 
business application and modernized payment system, the Veterans 
Service Network (VETSNET).
    OI&T also supports VBA's market research of business rules engine 
software, and other decision support technologies, which can be 
leveraged to support improved and expedited decision making, by claims 
processing personnel. OI&T recently released a Request for Information 
(RFI) from industry. The RFI resulted in the demonstration of 
technologies that may help support VBA's business strategy. The RFI 
process helps VA gain a better understanding of how private industry 
and other government agencies employ these technologies to support 
their business models. OI&T has also conducted analyses of technical 
architectures, business applications, and Commercial-Off-The-Shelf 
(COTS) products, utilized to support the business processes of the 
Social Security Administration (SSA), as well as the Veterans Affairs 
organizations of Australia and Canada.
    A Statement of Work (SOW) is currently being prepared to engage the 
services of a Lead Systems Integration Contractor (LSIC). The purpose 
of this contract is to assist VBA with the development of an 
overarching strategy and business requirements for the Paperless 
Delivery of Veterans' Benefits initiative. These key deliverables will 
enable OI&T to begin specifying the supporting technical architecture 
and business application.
    Mr. Chairman, I would now like to highlight how the utilization of 
data from the Veterans Health Information Systems and Technology 
Architecture (VistA) assists in the processing of disability claims. 
The business application used by VBA, to navigate and retrieve clinical 
data within VistA, is called the Compensation and Pension Record 
Interchange (CAPRI). Online access to medical data, housed in VHA's 
VistA, supports the disability benefits determination.
    CAPRI also provides access to some DoD medical records, through 
integration with the Federal Health Information Exchange (FHIE) 
framework. CAPRI was nationally deployed during Fiscal Year 2001, and 
delivers cutting edge ``point and click'' technology to the users' 
desktop. Since its deployment, the application has been repeatedly 
enhanced, as new categories of clinical data in VHA and DoD became 
available.
    Mr. Chairman, In closing I want to assure you that we remain 
steadfast in our efforts, to continuously optimize any and all 
information technology improvements, as we strive to improve our 
veterans' benefits IT environment. Our goal is that these efforts, 
coupled with the support of VBA and our partners in the private sector, 
will greatly improve the business processes, which will significantly 
enhance the disability claims process for our Nation's heroes. Thank 
you for your time and the opportunity to address these issues. I would 
be happy to answer any questions you may have.

                                 
     Statement of Raymond C. Kelley, National Legislative Director,
                       American Veterans (AMVETS)
    Mr. Chairman and Members of the Subcommittee:
    Thank you for providing AMVETS (American Veterans) the opportunity 
to submit our views regarding the use of artificial intelligence in VA 
claims processing.
    The claims backlog that plagues the Veterans Benefits 
Administration (VBA) has been a great concern for veterans, and AMVETS 
is pleased to see the Committee on Veterans' Affairs is taking the time 
to genuinely study this issue so long-lasting, effective changes can 
take place. AMVETS believes the use of artificial intelligence (AI) 
could greatly reduce the time involved in processing disability claims. 
The use of AI to reduce the amount of data provided to reflect only the 
information that is necessary to make timely and accurate decisions is 
not a new idea. The government has been using AI to select applicants 
for Federal positions and AI is being used in occupational health 
services to determine compensation for private and government 
facilities. So the question is not can AI be used to assist VA in the 
claims process, it is, to what extent can and should it be used?
    AMVETS generally supports the idea of using AI to improve the VA's 
claims processing system with some reservations. Integrating AI into 
the VBA to assist in the disability claims system could greatly improve 
the efficiency of the claims process or it could exacerbate the 
problems in the current method of adjudicating a claim. The use of AI 
would bring about a 100% electronic method of recordkeeping, which 
would provide easy access to records at all phases of the claims 
process. This would prevent the loss or misplacement of information.
    AI has the potential to significantly increase speed and accuracy 
during the triage and pre-determination stages of the claims process. 
At these points of the process, over and under development of claims 
happens all too often. With AI reviewing all of the incoming medical 
records, compensation and pension (C&P) exams that need to be ordered 
would be identified. This could prevent redundancy in examinations and 
ensure all necessary exams are requested and conducted, preventing the 
Rating Veteran Service Representative (RVSR) from beginning their work 
only to realize that additional C&P exams will be required. The rater 
now has the choice of either doing a partial rating, or stopping the 
rating, requesting the appropriate C&P exams, and completing the rating 
decision when the C&P exams return. Normally, this is a 60 day process.
    Disparity in the rating system from one regional office to another 
has become a major issue in disability claims compensation. The use of 
IA could, if developed properly, provide the rater with the appropriate 
diagnostic code for the determination of the percentage of disability 
for which a claimant qualifies. This is because AI also has the ability 
to recall all diseases and medical conditions and almost 
instantaneously compare all of the veterans' conditions against its 
database. Because of this, rare diseases and conditions that rating 
specialists see infrequently, such as keratoconus, will not be 
overlooked. Conversely, AMVETS has a concern that too much reliance on 
AI could result in not allowing logic to be used in determining the 
qualification of disability, something only a human can provide. So a 
line would need to be draw on how much responsibility will be given to 
the AI.
    However, the level of complexity of the AI would require brings 
about concerns on its own accuracy. The AI software would either have 
to be able to read hand-written documentation or physicians would have 
to type all of their findings. Also, there are certain tests that are 
represented in diagram or table form, such as the loss of field of 
vision test and the results of audiologist exams. The AI would need to 
understand and interpret these types of exams or be able to recognize 
their presence and prompt the rating officer to review them. This issue 
begs the question, will AI be grandfathered or will it have a starting 
point? If there is a grandfather provision, the AI will need to 
recognize medical terminology that has a tendency to change over time.
    To utilize artificial intelligence properly, the terminology 
inputted into the system would have to match the terminology the system 
recognizes. This brings to light the inconsistencies between the 
checklists contained in the Disability Evaluations Examination 
Worksheets used by VA physicians to conduct compensation and pension 
(C&P) physical examinations and the criteria contained in 38 C.F.R., 
Part 4, SCHEDULE FOR RATING DISABILITIES. In many cases, the 
descriptive words do not match. AMVETS' concerns are that AI software 
is not smart enough to provide accurate results when a physician's 
report does not exactly match the description of the rating even though 
the intent of the report describes the disability and the rating 
veteran service officer (RVSR) could interpret the meaning of the 
report though logic.
    There is a fundamental disconnect between the check list that 
prompts physicians on the Disability Evaluation Examination Worksheet 
that is used during the Compensation and Pension Exam, and the 
description in 38 CFR Part 4 that is used by the RVSR to determine the 
percentage, if any, that will be granted to a claimant. For AI to be an 
asset to the claimant, the physician's write-up must match the language 
used in 38 CFR Part 4. Under the current system of evaluation, a rating 
officer can determine the intent of a physician's evaluation.
    To receive a 30% disability rating for PTSD, 38 CFR Part 4 states:

          Occupational and social impairment with occasional decrease 
        in work efficiency and intermittent periods of inability to 
        perform occupational tasks (although generally functioning 
        satisfactorily, with routine behavior, self-care, and 
        conversation normal), due to such symptoms as: depressed mood, 
        anxiety, suspiciousness, panic attacks (weekly or less often), 
        chronic sleep impairment, mild memory loss (such as forgetting 
        names, directions, recent events).

    A claimant may have these conditions and they may be documented in 
the physician's evaluation. However, if the physician states the 
patient has some memory loss, rather than listing it as ``mild'' memory 
loss, then AI may not discern between the words or phrasing used by the 
physician and what it is looking for as it relates to 38 CFR Part 4. 
These logical word choices could easily result in the claimant 
receiving a lower or no rating for a disability. For artificial 
intelligence to be applicable it would be necessary to develop a 
Disability Evaluation Examination Worksheet that is based on the 
language used in 38 CFR Part 4 so semantics or synonyms would not 
reduce or prevent a claimant from receiving disability compensation. 
This would be easiest to facilitate on ratings of lost limbs or joint 
problems and become more complex when dealing with issues that have 
multiple variables such as PTSD or TBI.
    Again, AI could greatly benefit VA and the claimants if the system 
was allowed to work in all phases of the claims process and the 
information received by the system was completely and properly 
reviewed. Also, it is important to maintain the ability to intervene 
with human logic when necessary.
    Mr. Chairman, this concludes my testimony.

                                 
             Statement of Steve Smithson, Deputy Director,
    Veterans Affairs and Rehabilitation Commission, American Legion
    Mr. Chairman and Members of the Committee:
    Thank you for giving The American Legion the opportunity to present 
its views on the topic of using artificial intelligence to improve the 
Department of Veterans Affairs (VA) claims processing system. As VA's 
claims backlog continue to grow, it is important to explore new ways to 
utilize advances in technology that VA can implement to adjudicate 
benefits claims in a more timely and accurate manner. We commend the 
Subcommittee for holding this hearing.
    The American Legion welcomes innovative ideas, such as electronic 
claims processing and other uses of technology, which will enable VA to 
improve the service it provides to this Nation's veterans, especially 
in the arena of benefits delivery. We must, however, caution that 
automation does not guarantee quality claim development and speed does 
not guarantee accuracy or quality of data entry. Moreover, although the 
use of such technology might improve the process, it is not a magic 
bullet that will fix all the problems that are currently plaguing VA's 
disability claims process. Areas such as inadequate staffing levels, 
training, quality assurance, accountability, premature adjudication of 
claims and other problems resulting from VA's current work measurement 
system, as previously addressed by The American Legion in testimony 
before the Subcommittee, must be adequately dealt with before any real 
improvement resulting from use of artificial intelligence can be 
realized. Therefore, artificial intelligence based programs that direct 
the development and the adjudication of claims should be published in 
the Federal Register so that the public, especially stakeholders such 
as The American Legion, can provide written comments.
    The American Legion believes that the human element should never be 
removed from this equation and we are pleased that various experts that 
testified before the Subcommittee on the use of artificial intelligence 
in claims processing also agreed with this philosophy. Additionally, it 
must also be kept in mind that the bulk of the time and effort expended 
by VA in the disability claims process is not in the actual 
adjudication or decisionmaking part of the process; rather it is the 
part of the process that involves the development of the claim prior to 
adjudication. This process involves informing the claimant of the 
evidence that is needed to substantiate the claims as well as assisting 
the claimant in obtaining the needed evidence, such as military 
personnel and medical records, relevant medical evidence (both private 
and VA), scheduling compensation and pension examinations and other 
efforts necessary before the claim is ready to be adjudicated. Evidence 
development can be very time consuming and it is extremely important 
that any electronic claims system utilized by VA in the future 
adequately address this important part of the process, not just the 
actual adjudication of the claim, or any actual improvement in the 
current process will be minimal at best.
    Mr. Chairman and Members of the Subcommittee, I appreciate the 
opportunity to present The American Legion's view on this issue. This 
concludes my testimony.

                                 
   Statement of Kerry Baker, Associate National Legislative Director,
                       Disabled American Veterans
    Mr. Chairman and Members of the Subcommittee:
    I am pleased to submit for the record, the views of the Disabled 
American Veterans (DAV) on the issue under consideration today. In 
accordance with our congressional charter, the DAV's mission is to 
``advance the interests, and work for the betterment, of all wounded, 
injured, and disabled American veterans.'' We are therefore pleased to 
support various measures insofar as they fall within that scope.
    Regarding the implementation of an electronic claims' processing 
system, the DAV is not opposed to VA utilizing a test facility to begin 
implementation of such software on an experimental and limited basis. 
We will limit our support for this type of project in this manner until 
such time as adequate research and testing has been completed that 
provides evidence that a massive rollout of such technology is 
feasible. Additionally, while we support this novel idea, we caution 
Congress not to act in haste whereby legislation becomes law that 
imposes on the VA a requirement to implement such technology within a 
specific timeframe, especially when the potential technology has not 
been sufficiently identified. Imposing too short of a time limit for VA 
to implement such a virtual structure will only thwart its long-term 
success.
    A more reasonable approach would be to enact legislation that 
requires VA to submit to Congress a broad and over-arching plan by a 
reasonable date outlining the technology identified and the manner in 
which such technology will be utilized. Once this plan is complete, the 
groundwork will be laid for VA to coordinate with various entities, 
i.e., Congress, Veterans Service Organizations, Department of Defense, 
etc., in order to begin turning the plan into reality on a larger 
scale. The DAV would welcome the opportunity to work with the Agency, 
to include any contractors, in order to assist in the development of an 
electronic claims process system.
    The DAV also feels that using the term ``artificial intelligence'' 
is inappropriate as it is defined as ``the capability of a machine to 
imitate intelligent human behavior.'' \1\ The goal of any form of 
electronic claims process should be to automate, and thereby shorten as 
much as possible those portions of the claims process that currently 
consume the majority of time. Expecting a form of technology to imitate 
intelligent human behavior with respect to the decisionmaking process 
of VA's benefits delivery system, particularly where evidence weighing 
and judgment calls on such evidence are required, appears as an 
untenable goal--automation rather than human imitation is the first 
logical phase of this undertaking.
---------------------------------------------------------------------------
    \1\ Merriam-Webster's Collegiate Dictionary, 10th Ed., Pg 66., 
Merriam-Webster, Inc. Springfield, Massachusetts
---------------------------------------------------------------------------
    Contrary to some beliefs, the majority of time spent by VA on 
disability claims is in preparing the case for a decision. This 
includes receiving the claims by VA, establishing the claim in VA's 
current computer systems, and developing the evidence to support the 
claim. Evidence development, whether in the form of gathering military 
service records from the service department, military records from the 
Records Processing Center, private health records, VA health records, 
VA or private medical opinions, and stressor verification through the 
U.S. Army and Joint Services Records Research Center for claims of 
service-connection for Post Traumatic Stress Disorder, consumes the 
vast majority of the claims-processing time. Therefore, any viable 
electronic claims-processing system implemented with real expectations 
of shortening the claims process must focus on all VA functions and 
development leading up to the rating decision more so than just the 
rating decision itself.
    As far as automating functions of the decisionmaking process, 
caution must be exercised. Obviously, the first reasonable step in 
automating the final decisionmaking process would be to start with 
strict fact-based scenarios wherein a set of mandatory fact patterns 
equates to a mandatory award of benefits. The DAV believes that any 
attempt to go beyond this level of automation in the beginning phases 
of implementation, would likely cause more problems than it would 
solve. For example, rating decisions are required to contain adequate 
reasons and bases that explain to the claimant the purpose behind a 
particular decision.\2\ This becomes especially important when VA 
denies benefits. Without a detailed explanation of why a claimant is 
not entitled to a benefit sought, veterans and their dependents will 
have no recourse to correct what may only be minor deficiencies in 
their claims. Judgment necessary to communicate this type of 
explanation on a case-by-case basis will be inherently problematic for 
an automated process.
---------------------------------------------------------------------------
    \2\ 38 C.F.R. Sec. 3.103(b)(2006) (``Claimants . . . are entitled 
to notice of any decision made by VA affecting the payment of benefits. 
. . . Such notice shall clearly set forth the . . . the reason(s) for 
the decision. . . .'') (emphasis added).
---------------------------------------------------------------------------
    Additionally, any authorizing legislation concerning the issues 
herein must be accompanied by sufficient appropriations required to 
carry out such authorizations. To do otherwise would be tantamount to 
legislation incorporating its own veto.
    Ultimately, the DAV believes that proper utilization of technology 
has the potential for positive change by yielding the type of 
assistance that could bring the VA claims process into the 21st 
Century. Nonetheless, to be highly successful, Congress must avoid any 
proclivity to view this as a purely VA problem. The VA merely sits at 
the bottom of a whirlpool fueled by outside information. Automating the 
processes within the VA without automating the processes by which the 
VA acquires and assembles its information--information on which its end 
product is dependent--will do little to expedite VA's claims process.
    We appreciate the Committee's interest in these issues, and we 
appreciate the opportunity to present the DAV's views, which we hope 
will be helpful.

                                 
               Statement of Paralyzed Veterans of America
    Chairman Hall, Ranking Member Lamborn, members of the Subcommittee, 
Paralyzed Veterans of America (PVA) would like to thank you for the 
opportunity to present our views on this important issue. PVA 
appreciates the efforts of this Subcommittee to address the rapidly 
growing claims backlog.
    The interest to develop or adapt the necessary programs to help 
with claims processing is greater now, than ever before. In recent 
hearings Congress has informed the Department of Defense (DoD) and the 
Department of Veterans Affairs (VA) that they must accelerate their 
efforts toward achieving the long overdue policy of ``seamless 
transition.'' In testimony last year the information technology (IT) 
specialists reported that it will be at least 2012 before the DoD and 
VA medical records will be interoperable. Moreover, this projection of 
five years into the future does not take into account unanticipated 
problems. As both agencies work toward this monumental goal, we believe 
the VA should also continue the development of electronic processing of 
claims.
    PVA believes that the use of artificial intelligence (AI) in the 
processing of claims could help reduce the backlog of claims and 
increase the accuracy of the process. AI has been used in health care-
benefits delivery systems of the Federal Government and private 
industry for years.
    One serious problem recognized by the veterans' service 
organizations and other organizations that have examined the claims 
process is that many claims are not developed properly. A fully 
developed claim presented to the VA can generally be quickly decided. 
Using AI to process claims will not necessarily, overcome the problem 
of an unprepared claim, nor will it properly process an incomplete 
claim. We must acknowledge that the backlog problem starts in the 
beginning with proper development completing a claim. This is tedious 
and detailed work that must be completed by the veteran filing the 
claim for himself or herself, or working with a trained veterans' 
service officer. VA employees reported to the Veterans' Disability 
Benefits Commission that veterans who were assisted in filing a claim 
or appeal by either a VA benefits counselor or a veterans' service 
representative filed a better, well-documented claim.
    PVA Service Officers receive 18 months of training using electronic 
programs, text books, on-the-job observance and final testing. After 
completing this program the service officer is usually placed under the 
direction of a senior service officer. And yet, this only provides the 
basic knowledge necessary to properly prepare a claim. Still, there is 
little comparison between a PVA prepared claim and a claim prepared 
totally by the average veteran. We believe the VA must direct more 
effort toward proper preparation of a claim before it is submitted to 
the Rating Veterans Service Representative, (RVSR) for review.
    Many examples of the large disparity in compensation awards have 
been discussed in hearings during the 110th Congress. It has been 
reported that a significant percentage difference exists when comparing 
the same disability rating from two different VA Regional Offices 
(VARO) in separate states. The use of AI could standardize the final 
decision on compensation.
    The process of using AI could be integrated into the system in 2-4 
years as software is developed. However, it should be carefully tested 
and audited as the final product of the system is meant to be an 
accurately decided claim. Until that time, the VA will still need 
resources and staff. The Independent Budget for FY 2009 estimates the 
Comp and Pen service will need 12,184 total full time employees to 
address new claims and the claims backlog.
    We also recommend that the VA consider expanding the process of 
bringing back retired VA claims personnel. This has proven successful 
in expediting the processing of claims.
    PVA would favor testing the electronic processing of claims in 
certain regions perhaps as a pilot program. We would hope that VSOs, 
the veterans' community, and the interested parties would be involved 
in evaluating the finished product. If a program using artificial 
intelligence proves beneficial in accurately processing some types of 
claims, we believe this could be a significant step forward for the VA.
    PVA would like to thank you for the opportunity to submit this 
written statement. We appreciate the efforts of the Subcommittee to 
address the important issue of veterans' claims backlog. We look 
forward to working with the Subcommittee on this issue during the 
second half of this Congress. We would be happy to answer any questions 
that you might have.

                                 

          POST-HEARING QUESTIONS AND RESPONSES FOR THE RECORD

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Tom Mitchell, Ph.D.
School of Computer Science
Machine Learning Department
Carnegie Mellon University
5000 Forbes Ave.
Pittsburgh, PA 15213

Dear Dr. Mitchell:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if you could answer the enclosed hearing 
questions by the close of business on March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all full 
committee and subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
     The Use of Artificial Intelligence to Improve the VA's Claims 
                           Processing System
                         Tom M. Mitchell, Ph.D.
                 United States House of Representatives
                     Committee on Veterans' Affairs
        Subcommittee on Disability Assistance & Memorial Affairs
                      Hon. John J. Hall, Chairman

                               April 2008

    Question 1: If VA were using the technology you described, how many 
days could it take VA to rate a case given that some claims have upward 
of 10 conditions per claim?

    Response: The computer technology I describe executes very quickly 
on the computer, similar to the speed of TurboTax, which calculates 
taxes in a few seconds despite the complex Tax Code. Therefore, the 
number of days it would take the VA to rate a claim would be determined 
not by the computer technology (whose processing would only require 
seconds, not days), but instead by the delays it faces in collecting 
the data on which these decisions are based. In our round table 
discussion, the VA asserted that it faces large delays in pulling 
together the information required to rate a claim. Perhaps that delay 
could also be reduced by technology that maintains online a record of 
which information is in hand, and which is still to be obtained.

    Question 2: How complex is this type of technology and how long 
would it take to create such a system?

    Response: A prototype expert system, similar to TurboTax, could be 
implemented and ready for test deployment in a matter of months, not 
years, given a relatively small team of fewer than a dozen experts in 
this technology. Of course the full time-to-deployment includes more 
than the time needed to develop the software itself. It also includes 
the time it would take to make organizational changes to adopt the 
software, to train people on how to use it, etc. I do not have 
sufficient information to estimate the speed at which the VA could 
adopt the software once it was ready, so I will leave that part of the 
question to others.
    The timeline I would recommend for introducing such expert systems 
to the VA would involve (1) develop a prototype expert system to be 
used initially by only a small subset of the people who rate claims. 
This team of raters should work closely with the software team who are 
developing the expert system, to provide them with feedback about the 
user interface and functions, and to create a rapid re-design cycle in 
which the prototype software is developed, tested, refined, retested, 
refined again, etc. In this way, the VA users can have a strong voice 
in the design of the software that they will want in the end to use. 
(2) once a small team of VA raters have gone through this cycle, they 
can become the advocates to help spread this technology inside the VA, 
and to help train others on how to use it, and why it will be helpful. 
I expect the first step can be completed in about a year.

    Question 3: If VA could ``data mine'' as you have suggested, and 
reduce the mundane tasks staff currently perform, could the computer 
also send employees alerts to notify veterans of other benefits that 
they may not have applied for or may not be aware exist?

    Response: Yes, and in fact my contacts at Highmark Blue Cross tell 
me they use data mining of their own medical claims in precisely this 
fashion, for example to alert patients with chronic medical conditions 
of services that similar patients take advantage of, but which they are 
not filing for. The key idea here is this: data mining provides a 
computer-based approach to discovering regularities across many 
benefits claims. For example, one regularity might be ``if veterans who 
file for compensation for loss of a leg, typically also file for a 
wheelchair.'' Such a regularity can be automatically tested for each 
new claim, to determine whether it applies to this claim, and if so to 
alert the filer of the claim. Of course this is a simple rule, but the 
fact is that there can be hundreds or thousands of such regularities, 
and the data mining system can both discover these regularities and use 
them to send email alerts to veterans when relevant.

    Question 4: What would the cost be associated with implementing and 
operating such a system?

    Response: I am not certain of the answer to this. My advice would 
be to consult with one of the software companies that has already 
developed similar software for medical insurance companies. They would 
have much more expertise than I to answer this question.

    Question 5: You recommend that VA study the use of automation. What 
aspects of it do you foresee included in such a study? How long should 
such an analysis take?

    Response: I do recommend the VA seriously study the use of 
automation. It is clear that automation can be helpful, so I doubt that 
in and of itself is the right question. The appropriate question is 
instead ``which aspects of VA benefits can/should be automated or semi-
automated?'' Surely some simple steps (e.g., sending reminders to 
veterans that we are still waiting for information X to finalize the 
claim) can be automated. Other steps--those requiring substantial human 
judgment--might not be amenable to full automation. But even these 
steps in the workflow might be improved in accuracy and efficiency by 
computer assistance. One example is that when a human rater is 
considering a difficult case, the computer could retrieve and present 
the five most similar claims out of the many thousands that had been 
processed in the past. This is a use of computers to support human 
decisionmaking, and is what I referred to as case-based reasoning.
    More broadly, I believe the VA should consider introducing (1) 
fully electronic claims records coupled to the electronic medical 
reports for the veteran, (2) expert systems technology for automated 
and semi-automated processing of claims, (3) case-based reasoning 
systems to assist human claims processors by presenting the most 
similar historical claims, and (4) data mining systems that can 
discover regularities in historical claims and can spot new claims 
determinations that appear anomalous with respect to this historical 
precedent.
    In our round table discussion, I was surprised to find how little 
the VA has apparently studied this issue in the past, and even more 
surprised to see how little has come out of such ``studies.'' In my 
opinion, the kind of ``study'' the VA should do is not merely a study 
that produces a written document. Instead, it should do a study 
centered around active experiments that introduce computer technology 
in specific, limited, experimental ways, and the study should report on 
the experiments and the results. Such a study need not take years to 
complete--it should take months instead. Why not run a study that 
introduces each of the four technologies I suggest above, each in a 
limited way with a limited number of claims processors participating. 
If this were done, then a year from now we'd understand what really 
works, what really doesn't, and where the surprises lie. Studies that 
just involve thinking without experimenting are less likely to produce 
truly useful insights.

                                 

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Randolph Miller, Ph.D.
Professor, Former Chair
Department of Biomedical Informatics
EBL 416
Vanderbilt University School of Medicine
2209 Garland Ave.
Nashville, TN 37232-8340

Dear Dr. Miller:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if you could answer the enclosed hearing 
questions by the close of business on March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all full 
committee and subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
         Response to Questions from the Honorable John J. Hall
  Chairman, Subcommittee on Disability Assistance and Memorial Affairs
                     U.S. House of Representatives
  Hearing on the Use of Artificial Intelligence to Improve VA Claims 
                           Processing System
  From: Randolph A. Miller, MD of Vanderbilt University Medical Center
          Composed February 28, 2008 for March 6, 2008 Meeting
    Question 1: The VA does three things when it adjudicates a claim: 
1) it establishes there is a diagnosis; 2) it connects the diagnosis to 
the Veteran's time in the service; and, 3) it rates a level of severity 
of the disability. Do you think that the VA could use rule-based expert 
systems or some other application to complete these functions in 
assessing disability?

    Response: The short answer to the question is yes, the VA could use 
expert systems technology to improve the speed and quality of VBA 
disability claims processing. Nevertheless, the current, predominantly 
paper-based methods of claims processing must become more automated 
before it is even possible to consider use of expert systems. Such 
expert systems require accurate and detailed electronic records as 
input data in order to be effective.
    The testimony on January 29 before the House Subcommittee on 
Disability Assistance and Memorial Affairs clearly documented that 
paper forms the basis for current VBA claims processing. The hearing 
illustrated how paper-based records are easily misplaced or lost, and 
how they become fragmented as different groups within VBA require 
concurrent access to different parts of a Veteran's records while 
processing a claim.
    So the key question is how can the VA get from what is the current 
state of claims processing to a better, automated future state. In my 
written statement submitted for January 29, I outlined what I believe 
are three important phases of effort for doing so. Phase I would 
consist of building an electronic infrastructure to automate the steps 
of the existing disability determination system, in order to gain 
efficiency. While automation can eventually lead to a more complete 
redesign and overhaul of the claims processing system, it would be too 
disruptive--to both VA employees and to Veterans waiting for disability 
determinations--to scrap the current system and start over from square 
one all at once. Once the Phase I infrastructure was in place and being 
used gainfully, Phase II would enhance the automated infrastructure by 
adding decision support capabilities, some of which would involve 
expert systems technology. Clinical decision support programs provide 
assistance to healthcare workers in solving problems. Decision support 
includes expert systems as one of its many categories. The goal of 
decision support programs is to enhance the capabilities and efficiency 
of qualified and capable humans, not to replace the humans. Finally, 
Phase III would create quality monitoring and feedback processes to 
help the VA enhance, evolve, and improve disability determination 
processes over future years.
    The steps for each of the three phases are as follows:
PHASE ONE_BUILD INFRASTRUCTURE TO AUTOMATE STEPS OF EXISTING DISABILITY 
        DETERMINATION SYSTEM, TO GAIN EFFICIENCY
    Phase I, Step 1: At time of discharge from active duty, all 
Veterans should have a disability examination by a qualified healthcare 
practitioner that sets an initial, temporary level of disability. When 
any disability exists, appropriate payments should begin immediately 
upon discharge from the service as a prerequisite for discharge from 
active service.

    The record of the ``active service discharge disability 
determination'' should be stored electronically. When this disability 
determination occurs, there should be a target time, based on the level 
of disability and the Veteran's needs, to complete a more ``permanent'' 
disability rating (which would be equivalent in its objectives to 
current disability determination by VBA, but different process-wise, as 
detailed below). This goal ``time to `permanent' status determination'' 
should be set at active duty discharge to be one of 3 months, 6 months, 
or 1 year. Of course, disability status is not really ``permanent'' in 
that the disability condition can evolve over time, and the Veteran 
should always retain the right to appeal to assign a different status 
than was set. The Veteran would receive disability benefits at the 
temporary level until the ``permanent'' level evaluation is completed, 
and then receive benefits at the ``permanent'' level, with no 
interruptions in between.

    Phase I, Step 2: Identify the entire range of the types of records 
(documents and forms) that VBA must track to process an individual's 
claim. Give each record type a unique name, and create a definition and 
template for its contents. Similarly, identify the steps of disability 
determination that VBA goes through (i.e., each type of decision made 
by VBA) as it processes a claim from initial application to completion. 
Give a unique name to each step in processing.

    This is self-explanatory. The process should be completed within 2-
4 months, given adequate staffing and resources.
    See for example, prior work done within the VA in this regard: 
Brown SH, Lincoln M, Hardenbrook S, Petukhova ON, Rosenbloom ST, 
Carpenter P, Elkin P. Derivation and Evaluation of a Document-naming 
Nomenclature. J Am Med Inform Assoc. 2001;8(4):379-390.

    Phase I, Step 3: Develop and implement a plan to convert all 
relevant VBA records to electronic form.

    In changing to a more electronic disability determination system, 
one must be careful to convert almost all routine activities of VBA 
raters to being electronically based, with actions analogous to what 
they now do with paper. If both paper and electronic systems were in 
active use, a VBA rater would always have to check both systems to see 
if ``missing'' items in one system are actually not ``missing'', but 
present in the second system. It would potentially be worse--more 
cumbersome and slower for disability determinations--if VBA raters had 
to use both types of record systems together in processing a Veteran's 
application, than to use only one or the other system.
    The most straightforward way to begin conversion to electronic 
processing is to identify where paper records are currently generated, 
and where existing VHA/VBA software is applicable to creation of 
electronic versions of that paper-based information. For all other 
paper records that cannot be easily converted in this manner, the 
document naming system should be used to label them, and then records 
in each category should be electronically scanned to create 
electronically retrievable records with their dates of creation and 
their ``named'' labels. Ultimately, subsequent projects should develop 
tools to capture the content of all disability-related records at their 
source, through electronically-structured and standards-coded forms 
that create ``computer-processable'' content--using future electronic 
capture tools analogous to the existing VHA CAPRI system. The 
conversion should proceed based on each local/regional VBA office being 
converted as rapidly as possible before going on to the next office--
trying to do a VBA system-wide conversion on a single day could create 
too much chaos, due to the intensive level of support required in each 
local/regional office during such conversions. Once a large enough 
conversion support team existed, with adequate conversion experience, 
the pace of conversion could accelerate. All offices across the U.S.A. 
should be converted within a six to nine month span, using this model 
of starting slow and building momentum.

    Phase I, Step 4: Create a VBA automated record-tracking system that 
uses as a checklist the named set of steps created in Step 1 above, and 
uses the names of each record/document type to determine the status of 
each document for any given applicant.

    The checklists should be based on both an understanding of the 
definitions of each disability provided in CFA 38 Part 4 (and its 
amendments), and also an audit of adequately completed disability 
determination records for each type of disability possible in CFR 38 
Part 4. Thus, for each condition, there should be a list of the types 
of records and documents that are potentially relevant to disability 
determinations of that type, but also a list of the ``standard named'' 
processing steps that are required to carry the determination to 
completion. The record tracking system will show which steps and which 
documents are relevant in general, and which steps and which actions of 
that set the VBA has deemed required in a given Veteran's case, and 
then which of those required documents and steps are completed/
available through the electronic system, and which steps and procedures 
remain to be completed. The VA's AMIE and CAPRI systems have to some 
extent done this for components of the disability examination by the 
clinician, so the approach taken in developing those systems might 
inform the development of this more general tracking system.
    As noted above, once documents that comprise the full VBA record 
each have names, the existing documents initially can be scanned and 
stored and made reviewable by their formal names. Decisions that have 
been made would be displayed using their formal names. Scanning is a 
low-level form of technology that could free VBA workers from having 
only one paper document to review. However, scanning essentially 
creates a picture of the original document that humans can read and 
process. The latter scanned documents are, for the most part, not 
readable or understandable by machines.
    There are already portions of the VBA record that go beyond the 
simple technology of scanning, such as the CAPRI disability examination 
system used within VHA (and available for viewing at VBA). Note that 
scanning is a lowest-common-denominator first step that can move 
everything forward. The ultimate goal should be to make all disability 
related materials based on fully electronic data capture forms, like 
CAPRI.
    Once a document is scanned (or captured electronically using a 
template form), it should be stored centrally, with good backup and 
recovery capabilities, not at local offices on desktop computers--
modeled after the way that the VA's VistA electronic medical record 
system now works. The Internet can be used to view the centrally stored 
record from diverse VBA and VHA locations concurrently--substantially 
speeding up the ability to process a record. Similarly, the electronic 
checklist of steps required to complete a disability determination 
should be treated as if it were part of each Veteran's electronic 
medical record. The current status of each disability determination 
step, as well as target dates for completion of each required step, 
should be readily viewable from a central storage location. A reminder 
system can be built to warn VBA employees when the ``next step'' for 
each Veteran becomes due, and repeated reminders can be issued for 
steps that are overdue. At the completion of this step, there would be 
a tracking system in place that can immediately indicate where in the 
overall process each Veteran's application stands, and what documents 
are available to support both past and future decisions.

    Phase I, Step 5: Transition from a scan-based set of forms that 
were originally on paper to a computer-template based set of forms that 
collect primary data electronically (once, and from the most 
knowledgeable source) to enable better computer-based assistance in 
disability determinations.

    To develop electronic-templated data collection criteria for each 
disability condition would require human review of the latest version 
of CFR 38 and amendments for each condition, and creating: (a) a list 
of findings, coded in standard terminologies, such as SNOMED-CT or 
LOINC, required to be present to establish the disability, (b) a list 
of findings, coded in SNOMED-CT or LOINC required to be absent to 
establish the disability, (c) a list of findings that help to support 
the presence of the condition but which are not required to establish 
the condition, coded in SNOMED-CT or LOINC, (d) a list of the document 
types (using the standardized document names per above) that are 
relevant to determination of the specific disability condition, (e) the 
list of CAPRI frame identifiers that are relevant to determination of 
this specific disability, (f) names for each of the 700 conditions 
coded wherever possible in ICD, DSM, or SNOMED-CT, and (g) narrative 
text that describes the remaining criteria for the establishment of the 
specific disability condition that could not be coded in steps (a) 
through (c). This process could expand upon, and derive useful 
information from, the list of the required procedures/steps/documents/
records for each disability category that was built in Phase I, Step 4 
above.

    Phase I, Step 6: Create an automated ``dashboard'' for each VBA 
site (and which can be ``rolled up/summarized'' at the regional or 
national levels) that tracks all Veterans who have applied for 
disability compensation, the state of their applications, the status of 
their disability determinations, and uses color codes to indicate when 
steps are incomplete or overdue.

    This step is somewhat self-explanatory, and should be the end-
product of the previous steps in Phase I (especially steps 4 and 5). It 
should directly lead into Phase II Step 1.

    Phase I, Step 7: Determine how to best facilitate Veterans' 
application processes for disability determinations, using electronic 
entry forms and other automated methods to make the process more 
efficient and effective for Veterans and their families.
PHASE TWO_ENHANCE AUTOMATED INFRASTRUCTURE FOR DISABILITY DETERMINATION 
        WITH DECISION SUPPORT FEATURES
    Phase II, Step 1: Design, build, and implement, on top of the new 
VBA electronic tracking system, a rating system that enables claims to 
be electronically prioritized so that the most obvious or easiest to 
decide claims can be adjudicated more quickly.

    Once the automated tracking system is in place and running smoothly 
in pilot sites, a team of system designers, which should include 
computer-interface design experts, and socio-technical implementation 
experts, should obtain input both from expert disability raters (i.e., 
experienced, highly respected VBA employees) and from the most 
experienced and capable users of the new electronic tracking system. 
The collective wisdom of the group should focus on the question of how 
to determine, at any given time, which disability claim records are the 
most important to process rapidly (i.e., where a Veteran clearly and 
urgently needs assistance but is not getting assistance), and which 
records are easiest to process rapidly (i.e., require minimal effort to 
resolve).
    The result of these deliberations should be used to design and test 
a system that can create prioritizations that match those of human 
experts on preliminary system testing, with the goal of pointing out to 
VBA reviewers at each site what their priorities should be. Ongoing 
validation of such a system will be required, with frequent feedback 
from VBA end-users.
    The system should be merged with the tracking system of Phase I 
Step 6, to indicate which cases merit immediate attention of what sort, 
and with what priority. The system should set deadlines for relevant 
actions to occur, and remind VBA caseworkers of work requiring 
attention. An escalation process should inform managers at the local, 
then regional and national levels of the number and nature of cases 
requiring attention that have not received it in a timely manner, and 
could help to make sure adequate resources are devoted to resolving 
disability determinations promptly.
    The system should also be expanded to track the percent disability 
and the disability award amount for each disability category from CFR 
38 Part 4.
    The tracking system should also be made available to each Veteran 
who applies for disability determination to provide a summary level of 
where the application stands in its progress to completion, providing 
the Veteran with what steps have been completed, which steps remain, 
what documents have been received, and which are still outstanding. 
Care must be taken to make sure that the tracking system does not 
provide confusing or disturbing information to the applicant Veteran.

    Phase II, Step 2: Use expert diagnostic system techniques to assist 
human judgment in diagnosis of disabilities within VA rating system. As 
a part of this effort, completely redefine electronically the criteria 
for disability determination (in CFR 38 Part 4), and in doing so, 
create electronically actionable objective criterion lists as well as 
lists of which subjective judgments are required by human disability 
experts to complement the objective criteria in making decisions.

    A number of techniques developed over the past three decades for 
clinical decision support [see references at end of this step below] 
are relevant to future enhancements to a VHA/VBA disability 
determination and documentation system. At the national level, the VHA 
has been a major contributor to clinical decision support through its 
evolution of the VistA electronic medical record system. In addition, 
many talented individuals working within the VHA and VBA have also made 
contributions. It should be noted that Dr. Robert Kolodner, currently 
Director of ONC (the Office of the National Coordinator for Health 
Information Technology within the Executive Branch of the Federal 
Government) previously served with great distinction as a leader of 
informatics efforts within the VA.
    An important technological approach relevant to VBA claims 
processing is clinical diagnostic decision support applications [see 
references below for this step], which can be probabilistic (Bayesian), 
criterion-based, or heuristic (``artificial intelligence'' expert 
systems) in nature. In general, such systems take as input standardized 
vocabulary descriptors that characterize a patient's condition (such as 
history, physical examination, or laboratory findings) and produce as 
output a ranked list of possible diagnoses and a suggested approach to 
determining which diagnoses are present--with respect to a specified 
target set of allowed or possible diagnoses (which would be those 
specified by CFR 38 Part 4 and its amendments for the case of the VBA).
    It is important to note that diagnostic expert systems are useless 
in the absence of objective criteria to be used in making diagnoses. 
Such systems also require that all input data be in electronic form. 
The diagnostic criteria can include subjective ratings by human experts 
as part of the ``objective'' rating process, so long as the 
instructions for the human raters are clear, and the raters are 
calibrated to be equally reliable as other raters in carrying out such 
subjective assessments.
    Relevant to these considerations are two recommendations from the 
Institute of Medicine's (IOM's) June 2007 report, ``A 21st century 
System for Evaluating Veterans for Disability Benefits'' (National 
Academies of Science Press, 2007; Copyright  National Academy of 
Sciences. http://www.nap.edu/catalog/11885.html), listed below:

          ``IOM Recommendation 4-1. VA should immediately update the 
        current Rating Schedule, beginning with those body systems that 
        have gone the longest without a comprehensive update, and 
        devise a system for keeping it up to date. VA should 
        reestablish a disability advisory Committee to advise on 
        changes in the Rating Schedule.''
          ``IOM Recommendation 4-6. VA should determine the feasibility 
        of compensating for loss of quality of life by developing a 
        tool for measuring quality of life validly and reliably in the 
        veteran population, conducting research on the extent to which 
        the Rating Schedule already accounts for loss in quality of 
        life, and if it does not, developing a procedure for evaluating 
        and rating loss of quality of life of veterans with 
        disabilities.''

    The effort to redefine the conditions for which disability 
compensation is appropriate should be standards-based (ICD, DSM, 
SNOMED-CT, LOINC) as described above. Text-mining and natural language 
processing methods (see references for Phase II, Step 3 below) could be 
used to determine which coded terms are currently used in disability 
determinations through review of the thousands of existing electronic 
disability-related VistA and CAPRI records, and from samples of paper 
records converted by OCR (if of adequate quality) or alternatively by 
direct typing of samples of old records into electronic format. This 
review, coupled with the effort to extend disability criteria as 
recommended by the IOM Report, could result in computer-processable 
``criteria table'' definitions (see references 7-10 below for this 
step) for each disability condition that would maximize the objective 
representations of each condition (while still retaining free text if 
necessary to describe the aspects of human judgment required in each 
determination). As previously recommended, the list of document types 
and procedures relevant to determination of each disability category, 
as well as the orders required to carry out the procedures in VistA, 
could be added to an expanded revision of CAPRI.
    A key component of redefining the conditions listed in CFR 38 Part 
4 would be to include conditions that are well documented now that were 
not originally in the Codes, such as Post Traumatic Stress Disorder. 
Another important aspect would be to allow for coding for future 
disabilities that are clearly related to active service but which do 
not fit well into the existing code of disabilities at the time. Such 
``not elsewhere classified'' cases should be very carefully described 
in detail, and periodically reviewed (see Phase III, Step 3 below).
    Once the above representation scheme for each disability condition 
was in place, an expert system using the ``criteria table'' approach 
could be developed to assist VBA raters in determining the completion 
status of each disability determination, and added to a more advanced 
version of the previously mentioned dashboard system. The AI-RHEUM 
expert diagnostic system (references #7-10 below for this step), 
developed initially at the University of Missouri at Columbia and at 
Rutgers University, and subsequently at the U.S. National Library of 
Medicine (NLM), might be used as a starting point for the proposed VHA/
VBA expert system, because it uses ``criterion tables'' for diagnosis, 
which may fit well with a reformulated version of disability 
definitions in CFR 38 Part 4.
    Two of the original developers of that system are currently at 
NLM--Donald A.B. Lindberg, MD, is the Director of the NLM, and Lawrence 
C. Kingsland III, PhD, is a computer scientist at the Lister Hill 
Center within NLM.

    References for Phase II, Step 2:

     1.  Duda RO, Shortliffe EH. Expert Systems Research. Science. 1983 
Apr 15; 220(4594):261--268.
     2.  Miller RA, Pople HE Jr, Myers JD. INTERNIST1, An Experimental 
Computer-based Diagnostic Consultant for General Internal Medicine. N 
Engl J Med. 1982; 307:46876.
     3.  Bankowitz RA, McNeil MA, Challinor SM, Parker RC, Kapoor WN, 
Miller RA. A computer-assisted medical diagnostic consultation service: 
implementation and prospective evaluation of a prototype. Ann Intern 
Med. 1989; 110:82432.
     4.  Miller RA. Medical diagnostic decision support systems--past, 
present, and future: a threaded bibliography and brief commentary. J Am 
Med Inform Assoc. 1994;1(1):8-27.
     5.  Aliferis CF, Miller RA. On the heuristic nature of medical 
decision-support systems. Meth Inform Med. 1995 Mar;34(1-2):5-14.
     6.  Tom Mitchell. Machine Learning, McGraw-Hill: New York, NY. 
1997.
     7.  Kulikowski CA. Expert medical consultation systems. J Med 
Syst. 1983 Jun;7(3):229-34.
     8.  Kingsland LC 3rd, Lindberg DA, Sharp GC. Anatomy of a 
knowledge-based consultant system: AI/RHEUM. MD Comput. 1986 Sep-
Oct;3(5):18-26.
     9.  Porter JF, Kingsland LC 3rd, Lindberg DA, Shah I, Benge JM, 
Hazelwood SE, Kay DR, Homma M, Akizuki M, Takano M, et al. The AI/RHEUM 
knowledge-based computer consultant system in rheumatology. Performance 
in the diagnosis of 59 connective tissue disease patients from Japan. 
Arthritis Rheum. 1988 Feb;31(2):219-26.
    10.  Bernelot Moens HJ. Validation of the AI/RHEUM knowledge base 
with data from consecutive rheumatological outpatients. Methods Inf 
Med. 1992 Sep;31(3):175-81.

    Phase II, Step 3: Develop the ability to recognize, in clinical 
documents, terms that match findings or diagnoses of relevance to VBA 
disability determination.

    An important ``expert system'' technique relevant to clinical 
informatics is natural language text processing [see references at end 
of this step]. Using a target vocabulary of defined clinical terms or 
concepts, such as provided by the U.S. National Library of Medicine's 
Unified Medical Language System Metathesaurus, or by the SNOMED-CT 
terminology system officially endorsed by the U.S. Government (which is 
a subcomponent of UMLS), such programs can scan a ``free text'' 
document, such as a clinical note, and identify which of the target 
concepts are present in the document [references 1-4]. The utility of 
such an approach for VBA disability determination has already been 
demonstrated by a pilot project to identify spinal-injury-related 
findings from free text disability exam records, and to correlate those 
findings with an electronic representation of the criteria used by VBA 
to determine disability [reference 5].
    Ad hoc or heuristic approaches can combine manual techniques with 
semi-automated approaches to characterize clinical domains or 
conditions [references 6-7]. Such approaches have been used to derive a 
standardized vocabulary for patients' problem lists from a large set of 
examples in free text [reference 6], and to attempt to convert 
information stored in disparate DoD and VHA clinical record systems 
from one representation format to the other [reference 7].
    Once disability-relevant terms can be abstracted from clinical 
documents with adequate reliability, it will be possible to combine 
such lists with the diagnostic approaches mentioned in Phase II, Step 
2, to be able to suggest which disabilities that a Veteran (or possibly 
even an active servicemember prior to discharge from the service) might 
qualify and be screened for.

    References for Phase II, Step 3:

     1.  Uzuner O, Goldstein I, Luo Y, Kohane I. Patient Smoking Status 
from Medical Discharge Records. J Am Med Inform Assoc. 2008; 15: 14-24.
     2.  Chen ES, Hripcsak G, Xu H, Markatou M, Friedman C. Automated 
Acquisition of Disease--Drug Knowledge from Biomedical and Clinical 
Documents: An Initial Study. J. Am. Med. Inform. Assoc. 2008;15(1):87-
98. PrePrint published January 1, 2008; doi:10.1197/jamia.M2401
     3.  Friedman C, Shagina L, Lussier Y, Hripcsak G. Automated 
Encoding of Clinical Documents Based on Natural Language Processing. J. 
Am. Med. Inform. Assoc. 2004;11(5):392-402.
     4.  Denny JC, Smithers JD, Miller RA, Spickard A III. 
``Understanding'' Medical School Curriculum Content Using KnowledgeMap. 
J. Am. Med. Inform. Assoc. 2003;10(4):351-362. PrePrint published July 
1, 2003
     5.  Brown SH, Speroff T, Fielstein EM, Bauer BA, Wahner-Roedler 
DL, Greevy R, Elkin PL. eQuality: Electronic quality assessment from 
narrative clinical reports. Mayo Clin Proc. 2006; 81(11):1472-1481.
     6.  Brown S, Miller RA, Camp H, Giuse D, Walker H. Empirical 
Derivation of an Electronic Clinically Useful Problem Statement System. 
Ann Intern Med. 1999; 131(2):117-126.
     7.  Bouhaddou O, Warnekar P, Parrish F, Do N, Mandel J, Kilbourne 
J, Lincoln MJ. Exchange of Computable Patient Data Between the 
Department of Veterans Affairs (VA) and the Department of Defense 
(DoD): Terminology Standards Strategy. J Am Med Inform Assoc. e-
published Dec 20, 2007 doi:10.1197/jamia.M2498
PHASE THREE--CREATE A QUALITY FEEDBACK PROCESS TO ENHANCE AND EVOLVE 
        THE DISABILITY RATING PROCESS OVER TIME
    Phase III, Step 1: Using the disability determination tracking 
system described in Phase I, Step 6 and in Phase II, Step 1, develop 
progressively more sophisticated summary statistics that identify the 
following quality tracking metrics.

    Proposed quality tracking metrics include:

    a.  The time required from initial application submission by a 
Veteran to completion of disability status determination by VBA.
       1.  Subset by disability type (of the 700 conditions in CFR 38 
Part 4)
       2.  Subset by VBA Disability Office at local and regional levels
       3.  Subset by steps required from start to end, to find 
``logjams'' to work on
    b.  Use the data to determine which VBA raters fall outside the 
norms for the region or the Nation in terms of time to resolution of 
each Veteran's claim, consistency of VBA ratings, rate of responding to 
alerts and reminders, etc. These data should be used to help provide 
better follow-up training and instruction to VBA employees who require 
it, and if that is not effective, to find positions in VA for which the 
employee is better suited.
    c.  Use the data to review and refine the list of the core set of 
documents and steps required to determine each category of disability, 
over time. For example, at some time in the past, CT scans and MRI 
scans became more definitive and useful than plain x-rays for certain 
types of disability determination.
    d.  Consistency of assignment of disability category, percent 
disability, and awards for disability, for Veterans with similar sets 
of findings.

    Phase III, Step 2: Progressively refine the ability of natural 
language processing approaches to extracting disability-related 
information from free text records, and apply it to both DoD and VHA 
records to help screen for Veterans who may qualify for disabilities.

    This is somewhat self-explanatory, and represents time-wise 
refinement of Phase II

    Phase III, Step 3: Track disabilities that do not fit well into the 
existing disability code (see Phase II, Step 2 above), and based on 
prevalence (a) notify DoD and VA of their existence, and criteria for 
diagnosing them, and (b) once well-established, modify by amendment CFR 
38 Part 4 to include the new categories.

    Just as the medical profession tracks and defines a small number of 
new diagnoses annually (notable examples include Legionnaire's disease 
in 1976-77, AIDS in the early eighties, and Lyme Disease--which had 
been described and forgotten in the early 1900s, but re-emerged in an 
outbreak that started in Connecticut in the late seventies), there are 
new service-related disability conditions that will continue to emerge, 
such as Agent Orange exposure from the Vietnam War era, Gulf War 
syndrome from the Gulf War in the early nineties, and closed head 
injury brain trauma in the current Gulf War. It is important to have 
categories that allow assignment of disabilities at appropriate levels 
to Veterans even when their disorders do not fit neatly into existing 
CFR 38 Part 4 codes. It is equally important to encode the findings of 
such Veterans with ``in-between-the-codes'' disabilities in standard 
descriptive terminology so that such cases can be monitored at the 
national level, and when new patterns emerge, they can be further 
studied, categorized, and given new diagnostic labels.
    Development of new diagnostic labels for disabilities in Veterans 
has a number of ramifications. First, at the pragmatic level, 
definitions of the new conditions must be developed in the then-current 
electronic format used system-wide by the VHA and VBA, with 
appropriately coded standardized criteria and steps for diagnosis 
(consistent with Phase I and Phase II commentary above). Second, once 
such syndromes are characterized, if the conflict in which they arose 
is still ongoing, DoD personnel should be made aware of how to diagnose 
the condition in its early stages, so that hopefully the cause(s) can 
be determined, and preventive measures developed, whenever possible, 
within DoD, to lessen the occurrence of the condition during active 
service. Finally, once the condition has a formal definition, all VHA 
and VBA should be trained on how to detect and characterize the 
condition in a standard manner.

    Phase III, Step 4: The software and underlying medical knowledge 
underlying the VHA and VBA rating systems described in this document 
should be periodically reviewed with end-users and by outside experts 
for continuous quality-related improvements.

    Question 2: Can VBA claims be electronically prioritized so that 
the most obvious or easiest to decide claims can be adjudicated more 
quickly?

    Response: Yes, this was detailed in my answer to Question 1. Please 
refer to Phase II, Step 1 above.

    To recapitulate what was said above:

    Phase II, Step 1: Design, build, and implement, on top of the new 
VBA electronic tracking system, a rating system that enables claims to 
be electronically prioritized so the that most obvious or easiest to 
decide claims can be adjudicated more quickly.

    Once the automated tracking system is in place and running smoothly 
in pilot sites, a team of system designers, which should include 
computer-interface design experts, and sociotechnical implementation 
experts, should obtain input from both expert disability raters (i.e., 
long-term VBA employees) and from the most experienced and capable 
users of the new electronic system. The collective wisdom of the group 
should focus on the question of how to determine, at any given time, 
which disability claim records are the most important to process 
rapidly, and which records are easiest to process (i.e., require 
minimal effort to resolve rapidly). The result of these deliberations 
should be used to design and test a system that can do such 
prioritization, to point out to VBA reviewers at each site what their 
priorities should be. Ongoing validation of such a system will be 
required, with frequent feedback from VBA end users.

    Question 3: In your testimony, you mentioned a ``quality feedback 
loop.'' How can technology enhance the VA's efforts at monitoring 
quality in its claims processing?

    Response: In my answer to Question 1 above, items under ``Phase 
III'' all fall into the quality feedback category, and comprise a 
complete response to Question 3.

    Question 4: Can you further comment on the diagnostic system 
developed at the U.S. National Library of Medicine as to how it could 
be a starting point for VA disability claims?

    Response: This was discussed in the response to Question 1, under 
Phase II, Step 2. Lawrence (Larry) C. Kingsland, III, PhD, would be the 
most appropriate person to contact at NLM (301-496-9300), as he has a 
long history of developing and evolving criterion-based diagnostic 
systems.

    To recapitulate:

    Phase II, Step 2: (excerpt)

    Once the above representation scheme for each disability condition 
was in place, an expert system using the ``criteria table'' approach 
could be developed to assist VBA raters in determining the completion 
status of each disability determination, and added to a more advanced 
version of the previously mentioned dashboard system. The AI-RHEUM 
expert diagnostic system (references #7-10 for this step), developed 
initially at the University of Missouri at Columbia and at Rutgers 
University, and subsequently at the U.S. National Library of Medicine 
(NLM), might be used as a starting point for the proposed VHA/VBA 
expert system, because it uses ``criterion tables'' for diagnosis, 
which may fit well with a reformulated version of disability 
definitions in CFR 38 Part 4. Two of the original developers of that 
system are currently at NLM--Donald A.B. Lindberg, MD, is the Director 
of the NLM, and Lawrence C. Kingsland III, PhD, is a computer scientist 
at the Lister Hill Center within NLM.

                                 

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Marjie Shahani, M.D.
QTC Management, Inc.
1350 Valley Vista Dr.
Diamond Bar, CA 91765

Dear Dr. Shahani:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if you could answer the enclosed hearing 
questions by the close of business on March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all full 
committee and subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
                                                                QTC
                                                   Diamond Bar, CA.
                                                      March 3, 2008

Ms. Orfa Torres
335 Cannon House Office Building
Washington, DC 20515

RE: Questions on Artificial Intelligence to Improve the VA Claims 
Processing System

Ms. Torres:

    Attached is our response to the questions submitted by the 
Honorable John J. Hall regarding the Hearing on the use of artificial 
intelligence to improve the VA claims processing system. Please do not 
hesitate to contact me for additional information.

            Sincerely,
                                                 Dr. Marjie Shahani
                                                     Vice President
Enclosure
                               __________
              Response to Questions from Hon. John J. Hall
  Chairman, Subcommittee on Disability Assistance and Memorial Affairs
                     U.S. House of Representatives
   Hearing on the Use of Artificial Intelligence to Improve VA Claims
               Processing System held on January 29, 2008
                        From: Marjie Shahani, MD
            Senior Vice President, QTC Medical Service, Inc.
    Question 1: The Evidence Organizer sounds like it could make some 
impact on increasing the number of cases a day a rater could rate. A 
33% increase sounds good, but not great. Is there any way we could be 
doubling that figure?

    Response: Yes, the Evidence Organizer can make such an impact.
    Increasing the realized productivity of rating specialists to rate 
a case is dependent on multiple factors.
    First, it is critically important to move away from paper-based 
claims and supporting evidence. This would involve obtaining and 
converting claims and supporting medical evidence to a digital, 
searchable format. It is our understanding, from discussions with 
retired rating specialists and working with the VA, that the majority 
of their time is spent searching and weighing the medical evidence 
available on file. The Evidence Organizer will assist in accomplishing 
these tasks for the rating specialists thereby:

      Reducing the time the rating specialist spends reviewing 
and identifying the pertinent medical records for the individual issues 
claimed
      Reducing the time the rating specialist spends 
correlating the medical evidence to the diagnostic codes

    In addition, the Evidence Organizer can be linked to the current VA 
rating specialist's tool, RBA 2000, allowing flexibility of 
transferring information from the Evidence Organizer directly to the 
rating document.
    The Evidence Organizer will enhance the delivery of a more timely, 
standardized and accurate rating decision. Similar to other rule-based 
systems, utilization of the Evidence Organizer provides iterative 
improvements through fine-tuning of the rules and training the system 
thus producing increasing improvement of productivity and accuracy with 
use and time.

    Question 2: If a computer system could be matching key words as you 
described, would it still be necessary to generate a narrative report 
that needed to be read? Wouldn't it just be easier to avoid that step 
and let the computer match the criteria from the rating schedule to the 
findings in the exam template?

    Response: The Evidence Organizer does not generate a narrative 
report. It is a decision support system designed to enhance the 
efficiency of the Rating Veteran Service Representative (RVSR) or 
rating specialist by directly matching the medical evidence to criteria 
contained in the rating schedule.
    The Protocol Software application tool built by QTC and utilized by 
the examining physicians to perform the examinations according to VA 
AMIE worksheet requirements generates a narrative report. This is a 
separate tool all together.

    Question 3: To take the concept a step further. If an entire record 
is scanned into a computer using a high speed scanner and indexed at 
the same time--and that could be hundreds of pages in about half an 
hour--once that index was completed and key words were matched, then 
couldn't some sort of rule-based expert system as described by Doctors 
Mitchell and Miller apply the VA Rating Schedule and calculate a 
rating?

    Response: Yes. A rating decision could be calculated by a rule-
based expert system; however additional information would need to be 
added into the computerized system beyond the evidence and the rating 
schedule.
    The Rating Schedule and the Code of Federal Regulations embody the 
laws and procedures that apply to a properly evaluated disability. 
Based on our knowledge in working with the VA, these are not the only 
guides utilized by an RVSR in determining entitlement to benefits and 
selecting a percentage evaluation when certain disability criteria are 
met.
    Additional information that would need to be added to the expert 
system would include:

      Policy statements
      Procedure statements
      Administrative decisions
      Secretaries' decisions
      Court of Appeals for Veterans Claims (CAVC) precedents, 
and
      Other legal precedents governing Department of Veterans 
Affairs (VA)

    Incorporating all these into the rule-based expert system will 
allow for the software application to provide a recommended rating.
    The rating results generated by the expert system will be as 
accurate as the rules configured in it and the completeness of the data 
(files) accessible to it in digital searchable form.

    Question 4: We've often heard that claims have become more complex 
with over eight conditions per claim instead of just one or two. Could 
an expert system rate all of those conditions given that the claim was 
already in a ``ready to rate'' format such as the one described by Mr. 
Hunter? How long would it take?

    Response: Yes. A properly built and fine-tuned rule-based expert 
system could rate multiple conditions in a prompt manner. A specific 
timeframe to produce the completed decision would be dependent on the 
complexity of the algorithms and accuracy of the rules configured in 
the expert system and the infrastructure used to host the expert 
system; however, we anticipate this could be completed in minutes since 
we believe that the algorithms can be built in linear complexity.

                                 

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Mr. Ned Hunter
Stratizon Corporation
5425 Peachtree Way
Norcross, GA 30092

Dear Mr. Hunter:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if you could answer the enclosed hearing 
questions by the close of business on March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all Full 
Committee and Subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
                     Committee on Veterans' Affairs
       Subcommittee on Disability Assistance and Memorial Affairs
                             March 6, 2008
    ``The Use of Artificial Intelligence to Improve the VA's Claims 
                          Processing System''
    Ned M. Hunter, President and Chief Executive Officer, Stratizon 
                     Corporation, Atlanta, Georgia
    Question 1: Can you tell me more about where you are at with the 
Virginia pilot project?

    Response: The pilot is stalled due to funding. The application is 
built and ready to go.

    Question 2: Are other states or agencies using TurboVet and what 
have been their results?

    Response: No. Eight other states have expressed interest since 
presenting at NASDVA.

    Question 3: If VA was no longer relying on paper records and was 
primarily using an electronic system, how long do you think it would 
take them to process a claim?

    Response: It depends the integration of processes, both business 
and technical. We believe at least a 60% improvement in process cycle 
efficiency could be achieved.

                                 

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Mr. Gary Christopherson
Strategic Management and Performance
4327 Clagett Road
University Park, MD 20782

Dear Mr. Christopherson:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if you could answer the enclosed hearing 
questions by the close of business on March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all full 
committee and subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
 Hearing on Use of Artificial Intelligence to Improve the VA's Claims 
                           Processing System
        January 29, 2008 QFR Responses of Gary A. Christopherson
    Former Senior Advisor the Under Secretary for Health and Chief 
                          Information Officer
     Veterans Health Administration, Department of Veterans Affairs
    Former Principal Deputy Assistant Secretary for Health Affairs, 
                         Department of Defense
    Question 1: In your written testimony you state that you helped 
improve the VistA health information system. Can you emphasize your 
involvement? How is VistA a model and how can VBA implement this model 
or share in it?

    Response: Starting in the year 2000, I served as the Chief 
Information Officer for the Veterans Health Administration. When I 
arrived, the current plan by the previous CIO was to abandon VistA and 
move to a commercial product. After a thorough review of VistA from 
both the health care and information technology perspectives, I 
recommended that we build a new generation of VistA--called HealtheVet 
VistA--from the foundation of the existing VistA. That recommendation 
was unanimously approved by VHA, VA and OMB and was funded at a rate of 
an additional $125 million/year. That new development began in 2001.
    Based on internal reviews by VHA and the staff nationwide and on 
external reviews by a number of health care experts, VistA has been 
applauded as probably the most functional electronic health record 
system in the world. With next generation HealtheVet VistA, this 
functionality would be substantially enhanced and built on the latest 
information technology.
    VBA is using and could use it more effectively by cutting and 
pasting the health care information into the VBA applications. With a 
redesigned VBA application, the data could be directly imported without 
retyping or cutting/pasting. Importing the data greatly speeds up the 
search for the needed information, the application of that data to the 
VBA decisionmaking process, and the actual decision and any subsequent 
updates.
    The VistA application itself is a model for how one should 
construct the future VBA decisionmaking environment where the data is 
available electronically, in standard formats, coupled with decision 
support tools (e.g. artificial intelligence), and operating in real 
time supporting real time decisionmaking.

    Question 2: You mention in your testimony that claims can be 
processed in real time and rated in a single day. Given your background 
with VA and your knowledge of its capabilities, what does VBA need so 
it can capitalize on artificial intelligent technology and reach that 
real time goal?

    Response: First, the veteran should have the option of entering all 
the information they have online so that it is available to the VBA as 
soon as the application is submitted. Whatever medical or other 
documentation is available from other than DoD or VA can be submitted 
at attachments in standard formats. Whatever medical or other 
documentation is available from DoD or VA electronically should be 
available as it is generated. This is the case with VHA information 
today. It is partly true for DoD information today. As more of this 
data is digital and standardized, the power and speed of decision 
support tools increases.
    If more information is needed, a temporary rating could be issued 
based on what has been submitted and a signature by the veteran. If 
more information is needed, that should be able to be obtained within 
30 days and allow a ``permanent'' decision to be made within 30 days. 
For either, the appropriate payment should be started within 30 days of 
the application. Decision support tools (including artificial 
intelligence can easily support this model. This can be done today and 
can be done even better in the near future with enabling technology.

    Question 3: We have heard a lot today about what the technology can 
do to improve timeliness, accuracy, and consistency far beyond what 
people are capable of doing. So, what role does that leave for the 
Regional Office employees? If we free up their time from processing 
papers would they be able to spend more time advising veterans and 
their families on their claim and other available benefits or do you 
think implementing more technology will result in a loss of jobs?

    Response: Technology is unlikely to reduce the need for Regional 
Office employees any time in the near future. What it does do is allow 
them to be involved in more important work in support of the veteran 
and VA.
    First, there is always going to be the need for human intervention 
deal with more complex claims and where the decision support has limits 
(lack of digitized and standardized data; claims that fall outside 
current decision rules). Second, veterans need a partner to assist them 
prior to making a claim, when making a claim and on an ongoing basis as 
their situation changes. Third, veterans need people to ``meet and 
greet'' them at the door as a way to welcome them to a potential 
benefit based on their service to the Nation. Fourth, employees could 
become case managers where they actively and proactively engage with 
the veteran to make sure what needs to happen with respect to benefits 
does happen. This could reach further to helping coordinate, working 
with the veteran, VA programs with non-VA programs.
    There is no good reason why any existing employee should lose his/
her job. There is all the additional valuable work described above. 
Further, whatever loss of jobs does occur would likely occur over a 
long time period and could likely be handled through normal attrition.

                                 

                                     Committee on Veterans' Affairs
         Subcommittee on Disability Assistance and Memorial Affairs
                                                    Washington, DC.
                                                   February 4, 2008

Hon. James B. Peake, M.D.
Secretary
U.S. Department of Veterans Affairs
810 Vermont Avenue, NW
Washington, DC 20420

Dear Secretary Peake:

    In reference to our Subcommittee hearing on the Use of Artificial 
Intelligence to Improve the VA Claims Processing System on January 29, 
2008, I would appreciate it if Mr. Stephen Warren and Ms. Kim Graves 
could answer the enclosed hearing questions by the close of business on 
March 4, 2008.
    In an effort to reduce printing costs, the Committee on Veterans' 
Affairs, in cooperation with the Joint Committee on Printing, is 
implementing some formatting changes for materials for all full 
committee and subcommittee hearings. Therefore, it would be appreciated 
if you could provide your answers consecutively on letter size paper, 
single-spaced. In addition, please restate the question in its entirety 
before the answer.
    Due to the delay in receiving mail, please provide your response to 
Ms. Orfa Torres by fax at (202) 225-2034. If you have any questions, 
please call (202) 225-3608.

            Sincerely,
                                                       JOHN J. HALL
                                                           Chairman
                               __________
                        Questions for the Record
                  The Honorable John J. Hall, Chairman
                   House Veterans' Affairs Committee
       Subcommittee on Disability Assistance and Memorial Affairs
                            January 29, 2008
  Use of Artificial Intelligence to Improve the VA Claims Processing 
                                 System
Mr. Stephen Warren

    Question 1: In your statement you described a Request for 
Information that industry responded to that would help VA leverage 
decision support technologies, so what was the outcome of that process? 
How are you going to use that information?

    Response: Industry responses to the request for information (RFI) 
revealed a variety of tools and services available on the open market, 
which may assist the Department of Veterans Affairs (VA) in developing 
a business model that is less reliant on paper. During the vendor 
presentations stage of the RFI, it became evident that the paperless 
delivery of veterans benefits initiative has many interrelated 
elements, requiring a significant level of planning and integration to 
ensure successful implementation. To that end, VA has decided that the 
most prudent course of action is to engage the services of a lead 
systems integrator (LSI). The role of the LSI will be to work with VA 
to document business requirements, assess the current and planned 
information technology landscape, and develop an overarching program 
plan to implement the desired end-state. The information gathered 
during the RFI process will be provided to the LSI for consideration, 
as they assist VA in developing its implementation strategy.

    Question 2: VistA has helped bring the Veterans Health 
Administration to the cutting edge of healthcare in America. No other 
system has as dynamic an electronic system as VHA. So, why doesn't VBA 
benefit from it in the same way? Why are the Regional Offices taking an 
electronic medical record, printing it, and flagging it with post-its 
and rubber bands?

    Response: The Veterans Benefit Administration (VBA) does benefit 
from the advances in the Veterans Health Information Systems and 
Technology Architecture (VistA). Through the compensation and pension 
records interchange (CAPRI) interface, VBA personnel may request 
medical examinations necessary for the evaluation of disabilities 
claim. In addition, VBA personnel may search and download information 
contained in the VistA system, relevant to the veteran's claimed 
conditions. In our desired end-state, VBA employees will be able to 
capture this information electronically into the veteran's ``e-File'', 
for use in administering compensation benefits. This type of electronic 
information sharing and claims processing is a cornerstone of the 
paperless delivery of veterans benefits initiative.

    Question 3: During the hearing you were unaware of the disbanding 
of the Office of Seamless Transition. Why has that office disappeared 
while there are still so many problems that transitioning 
servicemembers and their families face? How is VBA informing veterans 
that they can file a claim online?

    Response: In October 2007, the Veterans Health Administration (VHA) 
reorganized the Office of Seamless Transition, in order to better meet 
the needs of severely wounded servicemembers and veterans. The clinical 
components of the office were realigned under the newly established 
Care Management and Social Work Service, whose mission is to address 
the needs of wounded and ill servicemembers. The employees, who were 
members of the Office of Seamless Transition, are a part of the new 
Care Management and Social Work Service.
    The transition assistance and case management that VA provides to 
Operation Enduring Freedom and Operation Iraqi Freedom (OEF/OIF) 
servicemembers, families and veterans have not changed as a result of 
the reorganization of the Office of Seamless Transition. It is VA 
policy that all claims for OEF/OIF veterans receive priority 
processing. The severely injured OEF/OIF claims are case managed, 
whether the seriously injured servicemember is transitioning from 
military to civilian life or remains on active duty. There is no higher 
priority for any VA employee, than ensuring the needs of those 
seriously injured in OEF/OIF are met in a timely and appropriate 
manner.
    VA uses a variety of methods to inform veterans that they can file 
a claim online. A few examples include face-to-face transition 
assistance program briefings; printed materials, such as pamphlets and 
brochures; and through our Web site under the heading ``Apply Online.''

    Question 4: Can you provided a more detailed report on the results 
of the RFI mentioned during the hearing? What was the industry response 
and what will VBA do with that information?

    Response: The RFI yielded 19 submissions from industry. These 
submissions covered such areas as business-rules engines, integration 
services, decision-support applications, and independent verification 
and validation services. Based on the initial submissions, VA requested 
that 11 vendors prepare a more targeted oral presentation. These 11 
vendors covered the spectrum of the categories described above. During 
the course of the presentations, it became evident that the paperless 
delivery of veterans benefits initiative has many interrelated 
elements, requiring a significant level of planning and integration to 
ensure successful implementation. To that end, VA has decided that the 
most prudent course of action is to engage the services of a lead LSI. 
The role of the LSI will be to work with VBA and the Office of 
Information and Technology (OIT) to document business requirements, 
assess the current and planned information technology landscape, and 
develop an overarching program plan to implement the desired end-state. 
The information gathered during the RFI process will be provided to the 
LSI for consideration, as they assist us in developing our 
implementation strategy.

    Question 5: Please provide a copy of the IBM Report requested 
during the hearing.

    Response: IBM will deliver its final report and provide executive 
briefings to VBA in March 2008. A copy of the final report will then be 
provided to the Committee.

Ms. Kim Graves

    Question 6: During the hearing, you mentioned an end-to-end claims 
process that will be paperless and computable, which based on the 
testimonies from the second panel is very realistic and should be the 
norm already. So, when will we see this new veteran's e-file and how 
long will this new system take to process a claim? Will it be more 
accurate and consistent?

    Response: The end-to-end claims process described in the testimony 
is our desired end-state vision for claims processing. We have already 
engaged in two pilot efforts which have demonstrated this type of 
``paperless'' claims process, using an ``e-File'', is feasible. The 
results of our RFI solicitation, as well as the presentations of the 
hearing panelists, provide for a number of potential opportunities to 
enhance the claims process. All potential solutions will be considered 
by the LSI as we develop our overarching strategy to reach the desired 
end-state. In addition, realistic performance metrics will be developed 
so that we may assess our progress in moving forward with this 
initiative. We fully expect to see improvements in the claims process 
as we move toward this end-state.

    Question 7: What is the problem with accepting an electronic 
signature?

    Response: VA agrees the claim process should be modernized to 
enable claimants to submit claims electronically and conduct other 
business with VA electronically, similar to the electronic business 
transactions that have become the mainstay of modern America. We are 
committed to this modernization as a high priority and are assessing 
the changes that are necessary to achieve this goal.

    Question 8: Why did it take VBA a year to rate and compensate Mr. 
Cleveland?

    Response: This information was provided to the Chairman only due to 
the personally identifying information contained in the response.

    Question 9: What can Congress do or what do we need to do to help 
VA to improve and use technology to process disability claims?

    Response: We believe we have identified a prudent course of action 
in seeking to bring on a LSI to assist us in developing and 
implementing our long-term vision of Paperless Delivery of Veterans 
Benefits. This strategy is consistent with the Social Security 
Administration's ``eDib'' initiative. As part of this process, we will 
more clearly define the end-state; we will articulate achievable 
milestones for implementation of capabilities and improved processes; 
and we will develop a long-term funding and acquisition strategy. 
Congressional support of our overall plan will be critical to our 
success.

                                 
