Rulebased Expert Systems

Rule-based expert systems may also find application in biosurveillance. A rule-based expert system (also known as a deterministic expert system) encodes medical knowledge in a set of if-then rules. Rule-based expert systems are simpler, conceptually, than probabilistic systems, but they are only appropriate for problems that do not involve reasoning under uncertainty. A rule-based formalism is appropriate when diagnostic or other knowledge can be represented as Boolean statements (e.g., if A and B then C).A hospital or biosurveillance organization can use a rule-based expert system to automate case detection when the findings are diagnostically precise (e.g., when case-detection is based on diagnostically precise data such as a positive tuberculosis culture [Miller et al., 1997]).

Figure 13.12 is an anthrax case detection rule from the Clinical Event Monitor, an embedded rule-based expert system developed by several of the authors (Wagner et al., 1997).

This rule is in a format that can be interpreted by a rule-based inference engine called CLIPS (National Aeronautics and Space Administration). The rule contains variables that represent findings of the disease anthrax. The part of the rule before the > symbol is the IF part and the part after the symbol is the THEN part. If all the conditions defined in the IF segment are true—wide mediastinum is found in a chest radiograph report and gram positive rods are reported in laboratory results within 10 days of each other—then the rule will conclude that there is a possible anthrax case. A typical rule-based expert system contains hundreds of such rules.

Was this article helpful?

0 0

Post a comment