Incremental induction of medical diagnostic rules

Abstract

This paper proposes a method for incremental updates of medical differential diagnosis, which consists of inclusive and exclusive rules. Since the addition of an example is classified into one of four possibilities, four patterns of an update of accuracy and coverage are observed, which give two important inequalities of accuracy and coverage for induction of probabilistic rules. By using these two inequalities, the proposed method classifies a set of formulae into four subrule layers. Inclusive rules will be updated by using these rule layers. However, since exclusive rules should cover all the examples of a decision, the update algorithm is implemented by enumerative operation for elementary attribute-value pairs. The proposed method was evaluated on datasets regarding headaches, meningitis and CVD, and the results show that the proposed method outperforms the conventional methods.