Title: The Effect of Inferring Work Location from Home Location in Performing Bayesian Biosurveillance
1The Effect of Inferring Work Location from Home
Location in Performing Bayesian Biosurveillance
- Chris Garman1, Weng-Keen Wong2, Gregory Cooper1
- 1RODS Laboratory, University of Pittsburgh,
2School of Electrical Engineering and Computer
Science, Oregon State University
Conclusions On our large scale anthrax attack
simulations, being able to infer the work zip
appears to improve detection time over just using
the home zip codes of the patients. Future work
involves performing more experiments and running
experiments over different sizes of simulated
anthrax attacks.
- Introduction
- Incorporating spatial information can improve the
performance of outbreak detection algorithms
(Buckeridge et. al. 2005) - Spatial and spatio-temporal detection algorithms
that monitor ED data typically use the patients
home zip code as an approximation to the point of
exposure - If attack occurs at work, the work zip code is a
much better approximation but the work zip code
is often missing
References Buckeridge DL. A method for
evaluating outbreak detection in public health
surveillance systems that use administrative data
Doctoral Dissertation. Stanford, CA
Biomedical Informatics, Stanford University
2005. Buckeridge, DL, Burkom H, Campbell M, Hogan
WR, Moore AW. Algorithms for rapid outbreak
detection a research synthesis. Biomed Inform
2005 38 (2)99-113. Cooper GF, Dash DH, Levander
JD, Wong WK, Hogan WR, Wagner MM. Bayesian
Biosurveillance of Disease Outbreaks. In
Proceedings of the Twentieth Conference on
Uncertainty in Artificial Intelligence. Banff,
Canada AUAI Press 2004. 94-103 p. Green MS,
Kaufman Z. Syndromic surveillance for early
location of bioterrorist incidents outside of
residential areas. In Proceedings of the
National Syndromic Surveillance Conference
CD-ROM. Boston, MA Fleetwood Multimedia,
Inc. 2004. Hogan WR, Cooper GF, Wallstrom, GL,
Wagner MM. The bayesian aerosol release
detector. In Proceedings of the National
Syndromic Surveillance Conference CD-ROM.
Boston, MA Fleetwood Multimedia, Inc. 2004.
- Methodology
- Estimate the probability P(Work Zip X Home
Zip Y) using historical data or census
information - Spatial detection algorithm now looks for a
specific spatial pattern in the home locations
and in the inferred work locations of the
patients. - This extension can be applied to any detection
algorithm that uses a probabilistic approach to
dealing with uncertainty - We applied it to the Population-wide Anomaly
Detection and Assessment (PANDA) algorithm
(Cooper et. al. 2004) - PANDA
- Models the effects of a large-scale airborne
release of inhalational anthrax on the population
using a causal Bayesian network - Population-wide approach each person in the
population is represented as a subnetwork in the
overall model
- Related Work
- (Green and Kaufmann 2004)
- Locate significant clusters of increased
morbidity in time and space using the patients
home addresses - Then finding the center of minimum distance to
these clusters - (Buckeridge 2005)
- Proposes two mobility models for estimating the
probability of an individual being in a spatial
unit given a time and home spatial unit - Workflow mobility model using workflow mobility
data from US Census - Non-workflow mobility model using estimates of
travel probability and trip distance
Acknowledgements This research was supported by
grants from the National Science Foundation
(IIS-0325581), the Department of Homeland
Security (F30602-01-2-0550), and the Pennsylvania
Department of Health (ME-01-737).
Anthrax Release
Global nodes
Location of Release
Time of Release
Angle of Release
Interface nodes
Person Model
Person
Person Model
Person Model
Person Model
models