Bayesian Network Anomaly Pattern Detection for Disease Outbreaks
W. Wong, A. Moore, G. Cooper, and M. Wagner. Proceedings of the Twentieth International Conference on Machine Learning, page 808-815. Menlo Park, California, AAAI Press, (August 2003)
Abstract
Early disease outbreak detection systems typically monitor health care data for irregularities by
comparing the distribution of recent data against a baseline distribution. Determining the baseline is
difficult due to the presence of different trends in health care data, such as trends caused by the day
of week and by seasonal variations in temperature and weather. Creating the baseline distribution
without taking these trends into account can lead to unacceptably high false positive counts and slow
detection times. This paper replaces the baseline method of (Wong et al., 2002) with a Bayesian network
which produces the baseline distribution by taking the joint distribution of the data and conditioning
on attributes that are responsible for the trends. We show that our algorithm, called WSARE 3.0, is
able to detect outbreaks in simulated data with almost the earliest possible detection time while
keeping a low false positive count. We also include the results of running WSARE 3.0 on real Emergency
Department data.
%0 Conference Paper
%1 wong03bayesian
%A Wong, Weng-Keen
%A Moore, Andrew
%A Cooper, Gregory
%A Wagner, Michael
%B Proceedings of the Twentieth International Conference on Machine Learning
%C Menlo Park, California
%D 2003
%E Fawcett, Tom
%E Mishra, Nina
%I AAAI Press
%K AnomalyDetection
%P 808-815
%T Bayesian Network Anomaly Pattern Detection for Disease Outbreaks
%U http://www.autonlab.org/autonweb/papers/y2003/14642.html?branch=1&language=2
%X Early disease outbreak detection systems typically monitor health care data for irregularities by
comparing the distribution of recent data against a baseline distribution. Determining the baseline is
difficult due to the presence of different trends in health care data, such as trends caused by the day
of week and by seasonal variations in temperature and weather. Creating the baseline distribution
without taking these trends into account can lead to unacceptably high false positive counts and slow
detection times. This paper replaces the baseline method of (Wong et al., 2002) with a Bayesian network
which produces the baseline distribution by taking the joint distribution of the data and conditioning
on attributes that are responsible for the trends. We show that our algorithm, called WSARE 3.0, is
able to detect outbreaks in simulated data with almost the earliest possible detection time while
keeping a low false positive count. We also include the results of running WSARE 3.0 on real Emergency
Department data.
@inproceedings{wong03bayesian,
abstract = {Early disease outbreak detection systems typically monitor health care data for irregularities by
comparing the distribution of recent data against a baseline distribution. Determining the baseline is
difficult due to the presence of different trends in health care data, such as trends caused by the day
of week and by seasonal variations in temperature and weather. Creating the baseline distribution
without taking these trends into account can lead to unacceptably high false positive counts and slow
detection times. This paper replaces the baseline method of (Wong et al., 2002) with a Bayesian network
which produces the baseline distribution by taking the joint distribution of the data and conditioning
on attributes that are responsible for the trends. We show that our algorithm, called WSARE 3.0, is
able to detect outbreaks in simulated data with almost the earliest possible detection time while
keeping a low false positive count. We also include the results of running WSARE 3.0 on real Emergency
Department data.},
added-at = {2006-09-19T17:26:09.000+0200},
address = {Menlo Park, California},
author = {Wong, Weng-Keen and Moore, Andrew and Cooper, Gregory and Wagner, Michael},
biburl = {https://www.bibsonomy.org/bibtex/2596f6931fa250949156c815ffca0ceab/kaixo},
booktitle = {Proceedings of the Twentieth International Conference on Machine Learning},
description = {Anomaly Detection},
editor = {Fawcett, Tom and Mishra, Nina},
interhash = {1ebbdee5c265fa08b7eada27f83035bd},
intrahash = {596f6931fa250949156c815ffca0ceab},
key = {wong2003b},
keywords = {AnomalyDetection},
month = {August},
pages = {808-815},
publisher = {AAAI Press},
timestamp = {2006-09-19T17:26:09.000+0200},
title = {Bayesian Network Anomaly Pattern Detection for Disease Outbreaks},
url = {http://www.autonlab.org/autonweb/papers/y2003/14642.html?branch=1&language=2},
year = 2003
}