The overall stated goal of this work is to advance the knowledge in the area of human risk assessment for developmental toxicants. The specific objective for this application is to develop a rule-based expert system for the hazard characterization and exposure estimation steps in the human risk assessment process. The purpose of such a system would be to assist individual counsellors in the human risk assessment area in their decision making process as they advise patients about developmental risks potentially incurred. The developed system will then be tested prospectively using scenarios of patient environmental and occupational exposure to compare its accuracy with that of nationally recognized human risk assessment experts. The primary method to be used for knowledge elicitation will be interview of the expert by a knowledge engineer. The """"""""rules-of thumb"""""""" derived from multiple national experts will be submitted to a delphi process to gain convergence or divergence on the specific rules. These rules will then be converted to a form suitable for knowledge representation in a computer program and one or more expert system models (as suggested by the need to test divergent rules) will be developed. Experts will be asked to characterize the hazard dose-response, mechanism of action, etc.) for known developmental toxicants and to estimate the hazard exposure for a set of predetermined patient exposure scenarios. Their performance will be compared with that of the computer system (and its programmed variations) to determine sensitivity, specificity, and positive and negative predictive value. Any areas of discrepant performance between the model(s) and the experts will be analyzed to determine methods to improve the expert system, as well as what relationships should further be subjected to experimental testing.
