Statistical Methods for Evaluating Exposure-Health Relationships

Nur H. Orak
Conventional experimental techniques are sometimes limited in their ability to assess the actual risk of chemical exposures. Therefore, there is a rising awareness of mathematical, computational, and statistical approaches to provide insight into the adverse effects of environmental contaminants. Richard Bach once wrote: "Any powerful idea is absolutely fascinating and absolutely useless until we choose to use it." Likewise, any data may be viewed as absolutely fascinating and absolutely
more » ... until we choose to understand and use it. Recent advances in science and technology provide alternative paths to develop effective risk-assessment methods for environmental contaminants. Moreover, these methods are more efficient in terms of time and cost. Therefore, I develop three Chapters to show the importance of statistical methods in environmental-health risk assessment, and highlight the potency of data-driven knowledge and multidisciplinary research for the future of environmental science and engineering. In Chapter 1, I review the potential risks of missing chemical data and concentration variability on mixture toxicity by developing 27 occurrence scenarios based on data from the literature. The @RISK software simulates random concentrations, assuming multivariate lognormal distributions for the mixture components. In Chapter 2, I demonstrate how a performance analysis can be implemented for a Bayesian Network (BN) representation of a dose-response relationship. I explore the effect of different sample sizes on predicting the strength of the relationship between true responses and true doses of environmental toxicants. In Chapter 3, I characterize the risk factors of a prenatal arsenic exposure network by using Bayesian Network (BN) modeling as a tool for health risk assessment.
doi:10.1184/r1/6723155.v1 fatcat:pkq6f5pdj5gybogejbox52dzua