A Probabilistic Bayesian Classifier Approach for Breast Cancer Diagnosis and Prognosis

Basically, medical diagnosis problems are the most effective component oftreatment policies. Recently, significant advances have been formed in medicaldiagnosis fields using data mining techniques. Data mining or Knowledge Discoveryis searching large databases to discover patterns and evaluate the probability ofnext occurrences. In this paper, Bayesian Classifier is used as a Non-linear datamining tool to determine the seriousness of breast cancer. The recordedobservations of the Fine Needle Aspiration (FNA) tests that are obtained at theUniversity of Wisconsin are considered as experimental data set in this research.The Tabu search algorithm for structural learning of bayesian classifier and Geniesimulator for parametric learning of bayesian classifier were used. Finally, theobtained results by the proposed model were compared with actual results. TheComparison process indicates that seriousness of the disease in 86.18% of cases areguessed very close to the actual values by proposed model.

Keywords:

Breast Cancer , Data Mining , Bayesian Classifier , Genie Simulator