|
Activity Number:
|
392
|
|
Type:
|
Invited
|
|
Date/Time:
|
Thursday, August 12, 2004 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Government Statistics
|
| Abstract - #300169 |
|
Title:
|
Using Simulation Methods to Assess the Validity of Aberration Detection Methods for Public Health Surveillance of Bioterrorism-associated Events
|
|
Author(s):
|
Lori Hutwagner*+ and Timothy Browne and Howard Burkom and Matthew Seeman
|
|
Companies:
|
Centers for Disease Control and Prevention and Centers for Disease Control and Prevention and Johns Hopkins University and Centers for Disease Control and Prevention
|
|
Address:
|
1600 Clifton Rd., Atlanta, GA, 30333,
|
|
Keywords:
|
simulations ; aberration detection ; biosurveillance
|
|
Abstract:
|
Empirical comparisons of aberration detection methods have been done. Simulated data are need for comparison of aberration detection methods based on actual sensitivity, specificity and time to detection. Datasets were generated using a negative binomial distribution based on the means and standard deviations from national, local, and syndromic surveillance. A given dataset includes daily data for six years with 1,000 iterations. A total of 10 types of outbreaks were randomly inserted though out the data. Eight of the 10 types of outbreaks are based on two types of outbreaks using two different means and two and three standard deviations. The other two types of outbreaks are spikes with the counts inflated by two and three standard deviations. A total of 55 datasets were generated. The simulated datasets are representative of actual data with varying seasonality, trends and frequencies. The outbreaks are smaller than and not as dramatic as some natural outbreaks appear. The outbreaks were left small so that they posed a challenge to all algorithms. Early detection of the outbreaks is of more interest.
|
