A Bayesian approach to aggregated chemical exposure assessment

Abstract

Human exposure to chemicals commonly arises from multiple sources, yet traditional assessments often treat these sources in isolation, overlooking their combined impact. We introduce a Bayesian framework for aggregated chemical exposure assessment that explicitly accounts for these intertwined pathways. By integrating diverse datasets - such as consumption surveys, demographics, chemical measurements, and market presence - our approach addresses typical data challenges, including missing values, limited sample sizes, and inconsistencies, while incorporating relevant prior knowledge. Through a simulation-based strategy that reflects the full spectrum of individual exposure scenarios, we derive robust, population-level estimates of aggregated exposure. We demonstrate the value of this method using titanium dioxide, a chemical found in foods, dietary supplements, medicines, and personal care products. By capturing the complexity of real-world exposures, this comprehensive Bayesian approach provides decision-makers with more reliable probabilistic estimates to inform public health policies.