UCLA

In industrial hygiene, prediction of a worker's exposure to chemical concentrations at the workplace is important for exposure management and prevention. The objective of this dissertation is to consider and address challenges in the statistical analyses of exposure data

in industrial hygiene. We outline a flexible Bayesian frameworks for parameter inference and exposure prediction. In particular, we will focus on two applications of the Bayesian approach on exposure data. The rst application is spatial interpolation of chemical concentrations at new locations when measurements are available from coastlines, as is the case in coastal clean-up operations in oil spills. We present novel yet simple methodology for analyzing spatial data that is observed over a coastline. We demonstrate four dierent models using two different representations of the coast. The four models were demonstrated on simulated data and two of them were also demonstrated on a dataset from the GuLF STUDY. Our contribution here is to oer practicing hygienists and exposure assessors with a simple and easy method to implement Bayesian hierarchical models for analyzing and interpolating coastal chemical concentrations.The second application is inference and prediction of chemical concentrations at the workplace using state space models. Exposure assessment models are deterministic models that are usually derived from physical-chemical laws that explain the workplace under theoretically ideal conditions. We propose Bayesian parametric and nonparametric approaches for modeling exposure data in industrial hygiene using a state space model framework which combines information from observations, physical processes and prior knowledge. Posterior inference is obtained via easy implementable Markov chain Monte Carlo (MCMC) algorithms. The performance of the dierent methods will be studied on computer-simulated and controlled laboratory-generated data. We will consider three commonly used occupational exposure physical models varying in complexity.