|
Activity Number:
|
87
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2004 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #300465 |
|
Title:
|
Multivariate Surveillance Schemes for Infectious Diseases on Multiple Locations
|
|
Author(s):
|
J. Michael Hardin*+ and Bo Hong
|
|
Companies:
|
University of Alabama and University of Alabama
|
|
Address:
|
Dept. of Information Systems, Statistics, and Management Science, Tuscaloosa, AL, 35487-0226,
|
|
Keywords:
|
multivariate ; surveillance ; infectious diseases ; control chart ; time-series ; spatial
|
|
Abstract:
|
A majority of the statistical research efforts on monitoring infectious disease surveillance have focused on univariate techniques, typically implemented on aggregated national disease data. However, most surveillance systems produce multivariate data: several reporting units are monitored simultaneously. The objective of this research is to design multivariate statistical surveillance schemes that simultaneously monitor incidence level of infectious diseases on multiple geographic locations, and provide early signals for the potential onset of epidemics in time and space. We examine multivariate control charts for autocorrelated processes by modeling the processes with multiple time-series method and then implementing multivariate control charts on forecast errors. Our proposed multivariate schemes outperform the univariate ones when applied to the weekly reported mortality of influenza and pneumonia from 1996 to 2004 for the United States. A systematic simulation study was also conducted to evaluate the performance of proposed surveillance schemes.
|
