Bayesian Approaches for Monitoring Influenza Surveillance Data

• Activity Number: 3
• Type: Invited
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Owen Devine*+ and Howard Burkom
• Companies: Centers for Disease Control and Prevention and The Johns Hopkins University
• Keywords: Influenza Like Illness ; Random Effects Models ; Poisson Mixtures ; Zero Inflated Poisson ; Over Dispersion ; Random Effects Models

Abstract

The Centers for Disease Control and Prevention (CDC) utilizes a variety of data sources for routine surveillance focused on both monitoring trends and timely detection of outbreaks of influenza-like illness (ILI). In this talk, we discuss the use of Bayesian methods to meet these surveillance goals. We address topics such as using posterior predictive distributions to assess the probability of an outbreak in recent surveyed information and the use of Poisson mixture models to reflect unknown combinations of underlying epidemic and non-epidemic states in time series of observed case counts. In addition, we discuss issues such as modeling over dispersion, temporal correlation, over abundance of zero case counts and decision rules for assessing outbreak probability that are ubiquitous in monitoring public health surveillance data. Applications of the methods are illustrated using data from CDC's ILINet surveillance system and we compare the Bayesian methods to frequentist approaches currently used to monitor these data.