Refinement of a Population-Based Bayesian Network for Fusion of Health Surveillance Data

摘要

The project was to refine a prototype population-based Bayes Network module for live implementation in the U.S. Department of Defense ESSENCE system to combine syndromic and clinical evidence sources to monitor health at hundreds of military care facilities. Evidence types included outpatient data records, laboratory tests, and filled prescription records. The multi-level approach included expanded data queries, data-sensitive algorithm selection, improved transformation of algorithm outputs to alert states, and hierarchical Bayesian Network training. Algorithmic and network thresholds were adjusted with stochastic optimization using 24 documented outbreak datasets.