Predictive Modeling of Cholera Outbreaks in Different Areas of Bangladesh

• Activity Number: 270
• Type: Contributed
• Date/Time: Monday, August 10, 2015 : 2:00 PM to 2:45 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Amanda Koepke* and Ira M. Longini and M. Elizabeth Halloran and Jon Wakefield and Vladimir Minin
• Companies: FHCRC and University of Florida and ASA and University of Washington and University of Washington
• Keywords: Bayesian ; Sequential Monte Carlo ; Hidden Markov Model ; Infectious diseases ; Time series

Abstract

While cholera is an endemic disease in Bangladesh, the exact relationship between seasonally varying environmental covariates and seasonal cholera outbreaks is unknown. We seek to develop a predictive model for the timing and magnitude of cholera outbreaks using environmental predictors. We use a Bayesian framework to fit a nonlinear dynamic model for cholera transmission in Bangladesh which incorporates environmental covariate effects. In this way we are able to estimate covariate effects in the context of a Susceptible-Infected-Recovered-Susceptible compartmental model of disease transmission, where these hidden states are only indirectly observed. To perform statistical inference under this hidden Markov model, we use recently developed particle filter Markov chain Monte Carlo methods. We analyze a unique data set consisting of cholera incidence data and environmental data collected from multiple geographical areas in rural Bangladesh over a period of sixteen years. We compare covariate effects and predictions in these different areas. Predictions from our model show a rise in the fraction of infected individuals weeks before the observed number of cholera cases increases.