Statistical Analysis and Applications of Physiologically-based Toxicokinetic Models in Environmental Risk Assessment

Abstract

Environmental risk assessment involves great uncertainty concerning the dose (exposure to toxic chemicals) and the response (risk or adverse health effect), and the relationship between the two. By simulating the route and metabolic mechanism of chemicals distributed within human body, physiologically-based toxicokinetic (PBTK) models have become increasingly important in dealing with these issues. However, many unknown model parameters and often limited data sources prohibit the popularity of PBTK models in practice. In this talk, I will introduce the standard PBTK models and discuss some related statistical issues. A Bayesian framework to estimate the unknown model parameters of the established state-space model is proposed. As an application, backward estimation of environmental exposure based on repeated blood measurements of sampled individuals is illustrated. Example and simulation studies will be given. Key Words: Bayesian; Exposure assessment; Markov Chain Monte Carlo simulation; Sample-important resample; Stochastic differential equation; State-space model