Automating the knowledge acquisition process in the construction of medical expert systems

In this paper, an inductive knowledge acquisition method that can be used as an aid for building certain medical expert systems is described. Given a collection of subjects that are described in terms of one or more attributes and are preclassified into a number of known classes (such as disease classes or classes that require certain type of therapy), this method is capable of detecting inherent probabilistic patterns in the data. Classificatory knowledge is then synthesized based on the detected patterns and made explicit in the form of classification rules. Based on these rules, the class membership of a subject can then be determined. The method has been implemented and tested with both simulated and real-world data. It has also been compared to some existing learning systems. The results show that the proposed learning method performs better both in terms of computational efficiency and classification accuracy.