Using genetic algorithms to search large, unstructured databases: the search for Desert Storm Syndrome

Exploratory data analysis problems have recently grown in importance due to the large magnitudes of data being collected by everything from satellites to supermarket scanners. This so-called "data glut" often precludes the effective processing of information for decision-making. These problems can be seen as search problems over massive unstructured spaces. A prototypical problem of this type involves the search, by Department of Defense medical agencies, for a so-called "Desert Storm Syndrome" which involves large amounts of medical data obtained over several years following the Persian Gulf conflict. This data ranges over more than 170 attributes, making the search problem over the attribute space a hard one. We propose the use of genetic algorithms for the attribute search problem, and intertwine it with search algorithms at the detailed data level. Computational results so far strongly suggest that our system has succeeded at the given tasks, requiring relatively few resources. They also have found no indication that a single syndrome or other medical entity is responsible for wide-spread adverse health ramifications among a significant cross-section of Persian Gulf War participants in the CCEP program. There are, however, numerous correlations of exposure/demographic information and associated symptoms/diagnoses which suggest that smaller groups may share common health conditions based on shared exposure to common health risk factors

Department dc:contributor.department

Information Technology Management

Grantor dc:publisher

Monterey, California. Naval Postgraduate School

Year dc:date.issued

1999

Author dc:creator

• Jacobson, David L.

Advisor dc:contributor.advisor

• Bhargava, Hemant K.

Language dc:language.iso

en_US

Handle dc:identifier.uri

https://hdl.handle.net/10945/9121

OAI identifier oai:identifier

oai:calhoun.nps.edu:10945/9121