Scientists "Mine" Data To Analyze Gulf War Illness

The article reviews a paper presented at an October INFORMS national meeting in Dallas by Hement K. Bhargava, Associate Professor of Information Technology at the Naval Postgraduate School, Monterey CA.

"Persian Gulf War Illness" is the name given to a collection of reported ill-defined and lingering symptoms and illnesses reportedly afflicting some U.S. participants in the 1991 Persian Gulf Conflict. Extensive medical evaluations, conducted under the Comprehensive Clinical Evaluation Program (CCEP), have not produced conclusive findings, but have left behind a large amount of data.

The GeneticStorms research involves exploratory analysis of the CCEP database, which contains data on about 20,000 U.S. troops. The data analysis perspective is that a syndrome, if there is one, will show up as an unusual pattern between combinations of attributes: as a group of diagnoses or symptoms shared by a cluster of participants sharing some common characteristics. Given that there are over 100 useful attributes in the database, the fundamental challenge is the number of data combinations (hundreds of millions or more). Operations research techniques are used to cope with this complexity.

The first two GeneticStorms studies found no syndrome affecting a large group, but identified smaller groups that share common health conditions based on shared exposure to common health risk factors (e.g. Participants reporting exposure to pesticides and non-Allied Forces food are 3.9 times more likely to be diagnosed with DJD/osteoarthritis and sleep apnea).

One particular analysis involved troops located within concentric circles around Khamisiyah, the site of an ammunition dump explosion that possibly caused exposure to chemical agents. This study determined that presence within 10km of the Khamisiyah blast was a significant factor for classification as worth a closer look by conventional and medical researchers.

The GeneticStorms project results were developed to guide medical investigators in further conventional and medical research. Specific recommendations include investigation of relationships between nerve gas, mustard gas, post-conflict reproductive problems, bleeding gums, weight loss, and muscle pain. Without the intensive data analysis, many aspects of so-called Gulf War Syndrome would likely remain a mystery.

The key contribution of GeneticStorms is a systematic, evolutionary search method to attack the problem. The entire methodology is encoded into a computer program.

The article appeared in the December edition of OR/MS Today, a publication of INFORMS. OR/MS Today can be viewed on-line at http://www.lionhrtpub.com/ORMS.shtml. Dr. Bhargava’s research can be viewed at http://bhargava.sm.nps.navy.mil/GSWeb/welcome.html.

The Institute for Operations Research and the Management Sciences (INFORMS) is an international scientific society with 12,000 members, including Nobel Prize laureates, dedicated to applying scientific methods to help improve decision-making, management, and operations. Members of INFORMS work primarily in business, government, and academia. They are represented in fields as diverse as airlines, health care, law enforcement, the military, the stock market, and telecommunications.