Decision assistance models for assessing occupational exposures

Outline of reasons and objectives
The objective of this study is to give companies a method for:
compiling various sources of information a priori making it possible to assess the levels of exposure to chemical pollutants, in particular statistical models and user expertise;
incorporating measurements conducted in companies in order to obtain final exposure diagnostic surveys taking into account not only the measurements but also other sources of information.
The end purpose is to optimise the number of measurements necessary for producing quality final diagnostic surveys.

Methodology
The statistical models developed previously using data from the COLCHIC database (measurements of occupational exposures to chemicals) were extended to a larger number of substances (26 substances). These models make it possible to estimate the occupational exposure on the basis of exposure determinants that are simple for the occupational safety and health specialist to collect (task, trade, number of staff in the company, sector of activity, type of substance, and substance). Then, other a priori information sources were added: expertise of the occupational safety and health specialist and mathematical models. A method of incorporating these information sources with measurements based on Bayesian updating was improved. Finally, a software prototype called TEXAS, implementing the methodology, was developed and tested.

Main results
In addition to the extension of the methodology and to two scientific publications, the main result is the development and validation of the TEXAS software prototype that constitutes evidence of the feasibility of implementing the method and of the possibility of making a tool available to occupational safety and health specialists.

Discussion
The COLCHIC database constitutes a mine of information on exposures that is absolutely exceptional at world level. Making the knowledge contained in COLCHIC available offers major scientific added value that also enables occupational safety and health specialists to have indications about the levels of exposure commonly encountered in the situations facing them. Occupational safety and health specialists can thus, as early as possible, adapt the metrological methods to be deployed and also adjust their decisions on the exposure diagnostic surveys (for non-regulatory purposes).
As a follow-on from this study, the “assessment by statistical models” part of TEXAS could be put on line in the form of a webpage on the INRS website. Subsequently, the other features of the TEXAS prototype could be gradually added, starting by the inputting of the exposure measurements and the possibility of conducting a regulatory diagnostic survey, which would lead to the current software, AltrexChimie, being replaced.