BBN-Based Decision Support for Health Risk Analysis

Abstract

Human health risk assessment (HRA) is the process to estimate the nature and probability of adverse health effects in humans. There are three issues about HRA: (1) hazard index (HI) cannot detail related diseases; moreover, we cannot exactly recognize the meaning of HI levels; (2) the same problems will also occur in human cancer risk; (3) the assumption of additive effect in human health for more than two toxic chemicals is problematic due to the violation of toxicology. To address these issues, this study proposes to use Bayesian belief networks (BBN) as a decision support for the higher level risk estimate which is able to represent the probabilistic relationships between all kinds of health effects and air pollutants. The study has several objectives: (1) to design the health risk procedure for BBN approach; (2) to represent the relationships between toxic substances, diseases and human effects with BBN; (3) to construct dose-response relationships by BBN; (4) to predict human disease and cancer risks due to specific toxic substance by BBN; and (5) case studies.