Dynamic Priors in Estimating Incidence and Prevalence in Epidemics

• Activity Number: 398
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Anisha Pulinchani Melethil*+ and Arni S.R. Srinivasa Rao
• Companies: Georgia Regents University and Georgia Regents University
• Keywords: Epidemic models ; Bayesian approach

Abstract

Incidence and Prevalence are one of the basic and fundamental observations recorded for understanding epidemic status in the population and progression of an epidemic. While the frequently used statistical techniques in estimating these two quantities are fairly understood well by the researchers, there still persists several uncertainties in epidemic predictions due to limitations in understanding the data and the process of generating such data. One such limitation is due to incomplete reporting of disease cases. The method of estimating the basic epidemic measures may be robust, but such robust methods heavily depend on selection of the right sample. One of the frequently used sampled data stratification could be by multinomial dynamic equations where size of the each sub-group of population varies by age. When age compartments are dynamic and a given disease is age dependent, the over all error of measuring incidence and prevalence for a population is weighted value of errors obtained for sub-groups. We try to understand these issues using Bayesian dynamical systems by introducing heterogeneity between sub-populations. There are two levels of prior information we will consider.