Skip to: top navigation | main navigation | main content

News & features

Tools developed to bring Bayesian methods to exposure assessment

A set of free tools could increase access to lognormal modelling for practising occupational hygienists conducting exposure assessments, its developers say.

Expostats is the most comprehensive, publicly available set of tools of its kind and "is meant to allow practitioners to utilise many theoretical developments formerly only available to academics", they add.

Many institutions providing occupational hygiene guidance recommend lognormal modelling for exposure assessment.

However, there are few tools for the technique suitable for practitioners, the scientists say in their paper, published in Annals of Work Exposures and Health late last year.

The set was developed by a group of scientists mostly based in Canada and led by Jérôme Lavoué, associate professor at the University of Montreal.

Expostats comprises four tools:

  • Tool 1 and its simplified version, Tool 1 Express, are for evaluating data from a group of workers who encountered very similar exposures;
  • Tool 2 is for evaluating the variability of exposure within a group and determining the probability of an individual having been overexposed; and
  • Tool 3 is for comparing exposure data across groups, which could be useful, for example, when evaluating the effect of an intervention.

Unlike other tools, they are based on Bayesian methods of statistical analysis which, according to the researchers, offer distinct advantages. They allow for quantitative integration of expert judgement and information from existing measurements.

They also provide direct probabilistic statements about the parameters of interest, which are easier for non-specialists to interpret than the alternatives, they say.

Finally, Bayesian methods deal well with "non-detects", data points that arise when, for the sake of conservatism, the limit of detection is recorded for a result lying beyond that limit.

Expostats is available in English and French and can be accessed through a web application, a standalone application or algorithms.

© CW Research Ltd. You may circulate web links to our articles, but you may not copy our articles in whole or in part without permission

CORRECTIONS: We strive for accuracy, but with deadline pressure, mistakes can happen. If you spot something, we want to know, please email us at: reportanerror@chemicalwatch.com
We also welcome YOUR NEWS: Send announcements to news@chemicalwatch.com