Activity monitoring: noticing interesting changes in behavior
KDD '99 Proceedings of the fifth ACM SIGKDD international conference on Knowledge discovery and data mining
Bayesian biosurveillance of disease outbreaks
UAI '04 Proceedings of the 20th conference on Uncertainty in artificial intelligence
A method for evaluating outbreak detection in public health surveillance systems that use administrative data
| Hi-index | 0.00 |
The Bayesian Aerosol Release Detector (BARD) is a biosurveillance system for detecting and characterizing disease outbreaks caused by aerosol releases of anthrax. A major challenge in modeling a population's exposure to aerosol anthrax is to accurately estimate the exposure level of each individual. In part, this challenge stems from the fact that the only spatial information routinely contained in the biosurveillance databases is the residential administrative unit (e.g., the home zip code of each case). To deal with this problem, nearly all anthrax biosurveillance systems, including BARD, assume that exposure to anthrax would occur at one's residential unit--a limiting assumption. We propose a refined version of BARD, called BARD-C, which incorporates the effect of commuting on a worker's exposure. We also present an experimental study to compare the performances of BARD and BARD-C on semi-synthetic outbreaks generated with an algorithm that also accounts for commuting.