- HyRisk – Hybrid Treatment of Uncertainties in Risk Assessments

Uncertainty is an essential aspect of risk assessments that should be accounted for in the analysis considering potential implications for the decision-making process. Uncertainty depends in particular on the information at hand regarding factors that influence risk. It is suggested in this report that different types of information, regarding variables that appear in risk assessment models, require different modes of representation of uncertainty. In the presence of " rich " information (for example measurements in significant numbers reflecting variability), a probabilistic mode of representation is applicable. On the other hand, when faced with " poorer " information (for example expert judgement), of an incomplete/imprecise nature (situation of partial ignorance) other modes of representation may seem preferable. When available information does not warrant a representation of uncertainty by single probability distributions, one alternative is to use families of probability distributions. This report illustrates the relation between families of probability distributions and so-called possibility distributions (also called " fuzzy intervals "). It presents the " hybrid " method for jointly propagating probabilistic and possibilistic information toward computing the uncertainty relative to the result of a risk assessment model. The theory underlying the hybrid method is presented in a didactic fashion. A learnware called HyRisk (downloadable free of charge at http://www.brgm.fr/hyrisk), that implements the hybrid method, is described and illustrated by an example with particular reference to health risk assessments. Disclaimer Neither the BRGM, nor any other organism, makes any warranty with respect to the HyRisk learnware, nor assumes any liabilities concerning: its use or misuse, the interpretation or misinterpretation of any results obtained with HyRisk, any damages resulting from using HyRisk.