Bayesian Analysis for Food-Safety Risk Assessment: Evaluation of Dose-Response Functions within WinBUGS

Abstract

Bayesian methods are becoming increasingly popular in the field of food-safety risk assessment. Risk assessment models often require the integration of a dose-response function over the distribution of all possible doses of a pathogen ingested with a specific food. This requires the evaluation of an integral for every sample for a Markov chain Monte Carlo analysis of a model. While many statistical software packages have functions that allow for the evaluation of the integral, this functionality is lacking in WinBUGS. A probabilistic model, that incorporates a novel numerical integration technique, is presented to facilitate the use of WinBUGS for food-safety risk assessments. The numerical integration technique is described in the context of a typical food-saftey risk assesment, some theoretical results are given, and a snippet of WinBUGS code is provided.