Fuzzy set and probabilistic techniques for health-risk analysis

摘要

A brief review of three non-fuzzy-set techniques available for performing risk analysis under uncertainty (probabilistic error propagation, Bayesian analysis, and the Shannon entropy) is given, using health-risk assessment to illustrate the approach. The advantages and difficulties in applying each of these techniques are pointed out. Next, a fourth technique is developed in some detail, namely, the fuzzy-set analysis of risk. Here, a given risk analysis problem evaluated under different sets of hypotheses may yield different fuzzy numbers, which can then be combined and compared. The elements of the methodology, which may be considered as an extension of interval analysis, are fuzzy numbers and operations, including their ranking, fuzzy regression to model a dose-response relationship, locating two-dimensional fuzzy numbers for risk management, and combining and ranking fuzzy risk estimates.