Disease Surveillance with Multiple Endpoints

• Activity Number: 530
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Yingqi Zhao*+ and Donglin Zeng and Amy H. Herring and David Richardson and Michael Kosorok
• Companies: The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill and The University of North Carolina at Chapel Hill
• Keywords: Disease Surveillance ; Meta-syndrome ; Multiple endpoints ; Multiple time series modeling

Abstract

Disease surveillance systems aim at rapid detection of disease outbreaks. Usually, the system can routinely obtain data on multiple endpoints from multiple data sources. Different observed data contain different information, and it will increase the certainty of identified outbreaks to use information from multiple data sources. We present an extension of univariate syndrome surveillance techniques. After first obtaining the predictions for each single endpoint series utilizing local spatiotemporal methods, we develop test statistics for the associated multiple residual time series based on Mahalanobis distance, with critical values determined with resampling. Moreover, we define a `meta-syndrome' constructed from principal component analysis based on endpoints studied, which can provide earlier indication of potential outbreaks by using the cross-correlation between different processes.