Monday February 24 2025   •   Time TBD

Environmental Statistics Day: “Mixture All Along: Statistical Methods for Estimating Complex Exposure-Response Functions”

With Howard Chang, PhD (Professor, Biostatistics and Bioinformatics, Rollins School of Public Health, Emory University)

Save the date! Registration and additional information coming soon.

Humans are simultaneously exposed to multiple correlated chemical and non-chemical environmental risk factors. Advancements in exposure assessments and statistical tools have enabled a shift towards studying the combined health impact of multiple exposures. This presentation will describe the use of approximate Gaussian process regression and Bayesian additive regression trees (BART) to flexibly characterize exposure-response functions. These approaches aim to address the computational challenges associated with Bayesian kernel machine regression (BKMR) and some parametric assumptions associated with quantile g-computation (qcomp). We apply these methods in several population-based epidemiologic studies to estimate health effects of ambient air pollution on emergency department visits and birth weight in Georgia, as well as an extension to identify heterogeneous health effects of heat waves. We will also describe challenges and approaches in visualizing and summarizing these complex exposure-response surfaces. Finally, we will discuss some practical and methodological issues related to the analysis of multiple environmental exposures.

Sponsored by: Michigan Center on Lifestage Environmental Exposures and Disease (M-LEEaD) Integrated Health Sciences Core