Dynamic Aggregation of Symptom Data.
Abstract
The health and well-being of U.S. service personnel assigned to duty in foreign countries is of vital importance to the successful completion of the intended missions. Upon relocation, such groups represent nonindigenous populations exposed to the endemic diseases of the region and other hazards associated with the assignment. Field epidemiologists and preventive medical specialists need ways to obtain and assess clinical data as quickly and accurately as possible and to relate the various signs and symptoms at the time of presentation into associated categories to support early detection of disease and illness. The objective of this study was to develop a computerized algorithm that would accept, sequentially in real time, patient symptom vectors and dynamically identity patterns of probable syndromes/diseases. The approach taken in this study was that provided by Adaptive Resonance Theory (ART). A simple ART algorithm was developed that was capable of sequentially accepting patient symptom vectors and dynamically clustering them into patterns or syndromes described by prototype symptom vectors. The basic ART algorithm was demonstrated with simulated symptom data from 300 patients representing 10 underlying syndromes. The algorithm was shown to correctly identify the syndromes and to cluster the patients into their correct syndromes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1996
- Accession Number
- ADA306139
Entities
People
- Amir Niknejad
- Daniel Pick
- John E. Angus
- Kevin Ames
- Robert Williamson
Organizations
- Naval Health Research Center