Hierarchical Bayesian tool for improving occupational exposure judgment

Occupational exposure assessments are often based on limited measurements and qualitative professional judgment, which can lead to substantial misclassification of exposure control categories (ECCs). Bayesian Decision Analysis (BDA) offers a way to integrate diverse information sources, but existing tools provide limited support for constructing custom priors. A web-based Exposure Decision Dashboard (EDD) was developed using a hierarchical Bayesian framework to combine semiquantitative checklists, mass balance models, and new sampling results into a common probability structure across American Industrial Hygiene Association (AIHA) ECCs. The EDD provides posterior probability distributions of the 95th percentile of the exposure distribution being located in each of the ECCs. Synthetic lognormal exposure datasets spanning a wide range of geometric means, geometric standard deviations, and X0.95/OEL ratios were used to evaluate performance under varying prior specifications, sample sizes, and variability, using accuracy, sensitivity, specificity, positive predictive value, and f1-scores for both binary (acceptable vs. unacceptable) and multiclass (five-category) classifications. Across clearly acceptable (ECC 0 to 2) and clearly unacceptable (ECC 4) scenarios, the Bayesian approach outperformed a frequentist X0.95-based strategy, even in the case of a uniform prior, reducing the probability of misclassifying unacceptable exposures as acceptable from 26% to 19% for a random draw of six samples. Informative priors that matched the true ECC further improved performance, but highly confident priors (maximum probability of 60% in a single category) produced unacceptable error rates when mis-specified. This prompted the selection of a 45% maximum probability value in any category as an optimized value that preserved high probability of correct classification while limiting the probability of incorrectly accepting clearly unacceptable scenarios to about 36%. Multiclass analyses showed overall accuracies above 80% across priors, with sensitivity and positive predictive value declining as prior confidence increased and performance improving consistently with additional monitoring data. The EDD (https://rhines.shinyapps.io/OH_Exposure_Dashboard/) provides occupational hygienists with an intuitive web-based tool to build and apply custom prior distributions from exposure determinants and monitoring data, closing a key implementation gap for Bayesian methods in exposure assessment. By capping prior probabilities in any category and visually highlighting conflicts between priors and likelihoods, the tool supports more accurate, conservative, and transparent decisions about exposure acceptability in line with the AIHA Exposure Assessment Strategy.

Authors

Institutions

Publication Details

Journal
Journal of Occupational and Environmental Hygiene
Published
2026-11-18
DOI
https://doi.org/10.1080/15459624.2026.2743747
Primary Topic
Occupational Health and Safety in Workplaces
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Hierarchical Bayesian tool for improving occupational exposure judgment

Gurumurthy Ramachandran, Ryan Hines
Journal of Occupational and Environmental Hygiene
Occupational Health and Safety in Workplaces
article

Hierarchical Bayesian tool for improving occupational exposure judgment

Gurumurthy Ramachandran, Ryan Hines
article en

Abstract

Occupational exposure assessments are often based on limited measurements and qualitative professional judgment, which can lead to substantial misclassification of exposure control categories (ECCs). Bayesian Decision Analysis (BDA) offers a way to integrate diverse information sources, but existing tools provide limited support for constructing custom priors. A web-based Exposure Decision Dashboard (EDD) was developed using a hierarchical Bayesian framework to combine semiquantitative checklists, mass balance models, and new sampling results into a common probability structure across American Industrial Hygiene Association (AIHA) ECCs. The EDD provides posterior probability distributions of the 95th percentile of the exposure distribution being located in each of the ECCs. Synthetic lognormal exposure datasets spanning a wide range of geometric means, geometric standard deviations, and X0.95/OEL ratios were used to evaluate performance under varying prior specifications, sample sizes, and variability, using accuracy, sensitivity, specificity, positive predictive value, and f1-scores for both binary (acceptable vs. unacceptable) and multiclass (five-category) classifications. Across clearly acceptable (ECC 0 to 2) and clearly unacceptable (ECC 4) scenarios, the Bayesian approach outperformed a frequentist X0.95-based strategy, even in the case of a uniform prior, reducing the probability of misclassifying unacceptable exposures as acceptable from 26% to 19% for a random draw of six samples. Informative priors that matched the true ECC further improved performance, but highly confident priors (maximum probability of 60% in a single category) produced unacceptable error rates when mis-specified. This prompted the selection of a 45% maximum probability value in any category as an optimized value that preserved high probability of correct classification while limiting the probability of incorrectly accepting clearly unacceptable scenarios to about 36%. Multiclass analyses showed overall accuracies above 80% across priors, with sensitivity and positive predictive value declining as prior confidence increased and performance improving consistently with additional monitoring data. The EDD (https://rhines.shinyapps.io/OH_Exposure_Dashboard/) provides occupational hygienists with an intuitive web-based tool to build and apply custom prior distributions from exposure determinants and monitoring data, closing a key implementation gap for Bayesian methods in exposure assessment. By capping prior probabilities in any category and visually highlighting conflicts between priors and likelihoods, the tool supports more accurate, conservative, and transparent decisions about exposure acceptability in line with the AIHA Exposure Assessment Strategy.

Journal of Occupational and Environmental Hygiene
Johns Hopkins University (US)
Johns Hopkins University, Centers for Disease Control and Prevention, National Institute of Environmental Health Sciences, National Institute for Occupational Safety and Health
Occupational Health and Safety in Workplaces
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon