Integrating Bayesian networks into fuzzy hypothesis testing problem - case based presentation

Bayesian networks have become a very popular model used to represent probabilistic medical knowledge bases. On the other hand the medical knowledge mostly contains fuzzy relationships among patients, diseases, symptoms and examination and assessment results. In this paper, we present a framework, which will enable us to assess the efficacy of treatment in the presence of imprecise and incomplete information. At its core is the intuitionistic fuzzy generalization of the McNemar test where Bayesian inference reasoning is employed to determine the membership and non-membership functions. Our approach integrates machine-learning techniques to support the hypothesis testing problem where the efficacy of treatment needs to be addressed with regard to imprecise and incomplete patient data stored in medical datasets.