Generating Bayesian networks from medical ontologies

Abstract

This work presents the development of a method for semiautomatically generating Bayesian networks for medical diagnosis and treatment from medical ontologies. Bibliographic review of the formalisms and related works are shown, as well as the formulated hypotheses, developed algorithm and the preparation of the evaluation experiment. Results pointed to correction rates of 82.1% and 90.5% for arcs and nodes generation, respectively.