You are here:
Consensus Modeling of Median Chemical Intake for the U.S. Population Based on Predictions of Exposure Pathways
Citation:
Ring, C., J. Arnot, D. Bennett, P. Egeghy, P. Fantke, L. Huang, K. Isaacs, O. Jolliet, K. Phillips, P. Price, H. Shin, J. Westgate, R. Setzer, AND J. Wambaugh. Consensus Modeling of Median Chemical Intake for the U.S. Population Based on Predictions of Exposure Pathways. ENVIRONMENTAL SCIENCE & TECHNOLOGY. American Chemical Society, Washington, DC, 53(2):719-732, (2019). https://doi.org/10.1021/acs.est.8b04056
Impact/Purpose:
This manuscript describes the third generation systematic empirical evaluation of models (SEEM) meta-analysis. A collaboration of exposure researchers has developed databases and mathematical models allowing for high-throughput exposure (HTE) forecasting. These "exposure predictors" (data and models) been grouped into four pathways (residential, dietary, pesticidal, and industrial). All exposure predictors have been evaluated and calibrated via Bayesian multivariate regression using human intake rates inferred for 114 chemicals from a large bio-monitoring survey. Machine learning models based on chemical structure and physico-chemical properties predict whether or not each pathway is relevant to a library of over 680,000 chemicals, allowing an exposure estimate for each chemical based on the calibrated predictors
Description:
Prioritizing the risk posed to human health by chemicals requires tools that can estimate exposure from limited information. We consider an “exposure pathway” to include a chemical source, interaction with the environment, and a receptor (such as a person). Here we present a calibrated, consensus model based on 14 models and other exposure predictors. Machine learning predicts the probability that a chemical might be associated with exposure via four different example pathways: near-field (residential), dietary, far-field industrial, and far-field pesticide. On a pathway basis, we examine 114 inferred chemical intake rates for the median U.S. population. WE find that the chemical exposure by the residential pathway is much higher, and that the dietary and pesticidal pathways are both slightly higher than the overall average intake rate. We evaluate the ability of the predictors to augment the pathway averages (higher or lower) and find that we can explain roughly 85% of the chemical-to-chemical variance. We extrapolate to other chemicals to predict relevant pathway(s) and median intake rate for 687,359 chemicals. We finding that 87% of these chemicals have less than 80% chance of exposure via any of the four modeled pathways. This approach separates 1033 chemicals with some probability of median population intake rates exceeding 1 mg/kg BW/day from the other 686,326 chemicals where there is 95% confidence that intake rate would be lower than that.
URLs/Downloads:
DOI: Consensus Modeling of Median Chemical Intake for the U.S. Population Based on Predictions of Exposure Pathways