Expert System for Stroke Classification Using Naive Bayes Classifier and Certainty Factor as Diagnosis Supporting Device

Abstract

Stroke is a disease that has high mortality rate in Indonesia and worldwide. This disease is dangerous if not immediately treated because the brain circulatory disorder can cause a permanent disability or death. Diagnosis process of stroke disease can be assisted by an expert system using Naive Bayes Classifier and Certainty Factor. Decision making by Naive Bayes Classifier uses total probability value of all the diagnostic criteria on the existing database. Certainty Factor uses a weight value combination from measure of believe provided by expert. This expert system research aims to assist in the early diagnosis of a neurologist in diagnosing potential stroke patient from several diagnostic criteria as well as determine the accuracy of the expert system program created. The sample used in this study are the hospital medical records of 130 patients from Dr. Soetomo Hospital of Surabaya consists of 80 stroke patients data and 50 non-stroke patients data. Expert system program uses 12 category for diagnosis as an input and two statement outputs between stroke or non-stroke. Based on the analysis from the 25 testing data of 105 training data we obtained the accuracy of the method Naive Bayes classifier by 96% and of the method of Certainty Factor of 84%. Expert system program using Naive Bayes Classifier and Certainty Factor can be a device for stroke diagnosis.