Process-Based Bayesian Melding of Occupational Exposure Models and Industrial Workplace Data

• Activity Number: 597
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Sudipto Banerjee*+ and Joao Vitor Dias Monteiro and Gurumurthy Ramachandran
• Companies: University of Minnesota and University of Minnesota and University of Minnesota
• Keywords: Occupational exposure models ; Bayesian melding ; Gaussian process

Abstract

A primary issue in industrial hygiene is the estimation of a worker's exposure to chemical, physical and biological agents. Mathematical models, based upon physical principles, are being increasingly used as a method for assessing occupational exposures. However, predicting exposure in real workplace settings is constrained by lack of quantitative knowledge of exposure determinants. We recognize these limitations and provide a rich and computationally effective Bayesian hierarchical framework that melds the physical model with the observed workplace data. We reckon that the physical model, by itself, is inadequate for enhanced inferential performance and deploy multivariate Gaussian processes to capture extraneous uncertainties and underlying associations. We formulate two different approaches: (i) where the inputs to the physical model are unknown and must be estimated, and (ii) where the inputs are controlled and we seek to meld the output from the physical model with the observed data.