Bayesian Hierarchical Spatial Modeling: An Application in Modeling the Prevalence of Brucellosis Among Elk in Yellowstone Ecosystem

• Activity Number: 348
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Pritam Gupta*+ and Timothy Robinson and Richard Anderson-Sprecher
• Companies: University of Wyoming and University of Wyoming and University of Wyoming
• Keywords: Brucellosis ; Bayesian hierarchical spatial model ; Poisson-Gamma mixture ; spatial autocorrelation ; residual spatial structure ; spatial epidemiology

Abstract

Brucellosis is a bacterial disease that causes fetal abortions, decreased milk production and infertility in ungulates (e.g., elk, cattle). The Greater Yellowstone Ecosystem (GYE) is the last reservoir of this disease in the United States. The Wyoming Game & Fish Department maintains an elk vaccination program to protect both wild elk and domestic cattle. Our main objective is to provide space-time predictions of Brucellosis prevalence and related quantities to aid decisions regarding the timing and location of vaccination efforts. Towards this objective, we report on our efforts to build a Bayesian hierarchical spatial model that monitors the changes in residual spatial structure of the number of infected elk. This model is subsequently used for exploring unknown risk factors. The proposed model is based on a Poisson-Gamma mixture and it properly accounts for observed region-level covariates and unobserved spatial autocorrelation. The monitoring of the residual spatial structure is helpful in identifying those covariates which are potential risk factors for Brucellosis prevalence.