This article proposes defining focal elements in the Dempster-Shafer theory as fuzzy sets in an application to medical diagnosis support. Membership functions for medical parameters of "fuzzy" nature are constructed. A diagnosis support consists of Bel measure calculation only for these focal elements that have membership function values grater than a "truth" threshold. Coherence between membership function shapes and the truth threshold is shown and a new way of membership function designing is proposed. An extension of the "truth" threshold for nonfuzzy focal elements is proposed that make a unification of symptoms interpretation during diagnosis support possible.
The paper presents a study on data-driven diagnostic rules, which are easy to interpret by human experts. To this end, the Dempster-Shafer theory extended for fuzzy focal elements is used. Premises of the rules (fuzzy focal elements) are provided by membership functions which shapes are changing according to input symptoms. The main aim of the present study is to evaluate common membership function shapes and to introduce a rule elimination algorithm. Proposed methods are first illustrated with the popular Iris data set. Next experiments with five medical benchmark databases are performed. Results of the experiments show that various membership function shapes provide different inference efficiency but the extracted rule sets are close to each other. Thus indications for determining rules with possible heuristic interpretation can be formulated.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.