You are here:
High-throughput exposure models for critical pathways: Implementation and parameterization of models for occupational exposure
Citation:
Phillips, K. High-throughput exposure models for critical pathways: Implementation and parameterization of models for occupational exposure. U.S. Environmental Protection Agency, Washington, DC, 2021.
Impact/Purpose:
This product delivers a model that uses chemical structure and knowledge of industrial sectors in which that chemical is used to predict 1) if a chemical would likely be detected in that indstrial sector and 2) if it were to be detected, what would be the concentration in the air in that workplace at which it would be detected. This model can perform predictions for thousands of industrial sector - chemical pairs rapidly, allowing for rapid estimation of concentrations to aid in chemical prioritization in assessments.
Description:
With the passing of the Frank R. Lautenberg Chemical Safety for the 21st Century Act, the Toxic Substances Control Act (TSCA) was amended to require risk assessors to consider susceptible or highly exposed subpopulations when prioritizing compounds for risk assessment. While high-throughput consumer exposure models can be modified to account for some populations that can be considered susceptible (e.g., sensitive age groups), there are relatively few high-throughput models capable of addressing the chemical exposure scenarios encountered by workers (a highly exposed population). Further, existing low-throughput models for occupational exposure pathways and routes are available, however these models cannot be run for thousands of chemicals in a reasonable time frame. This product consists of a Bayesian hierarchical model whose aim is to create a flexible statistical framework that can be broadly applied to any substance (if its physicochemical properties and industrial sectors in which it is used can be estimated). The data used both to train and test this model were measured air concentration values from OSHA’s Chemical Exposure Health Data. The processed data used in this model were collected from 1984 (the first year that data were reported by OSHA) until 2018 (the last year that data were reported at time of initial modeling effort). A high-throughput model has been developed, parameterized, and evaluated to predict detection of chemical substances in various industries and, if detected, the air concentration of that substance is also predicted. Good agreement between model predictions and measured air concentration values are produced from this model. Predictions from these models will directly feed new Occupational ExpoCast/SEEM consensus models, which are a critical component of proposed ORD/OPPT workflows to support TSCA prioritization candidate identification. This model may also be suitable for broadly screening out high risk substances for further study.