A support vector machine approach for predicting heart conditions

Early diagnosis of heart-related problems can potentially reduce mortality rate and help patients maintain a better quality of life. Recently, advanced mathematical and statistical models enable us to analyze the continuous flow of data and to predict hazardous heart conditions. In this paper, we present a categorical feature-based classification method using support vector machines (SVMs) to predict the heart condition. Existing medical and demographic data of the patients will be used as input for classifying the heart condition as normal or abnormal. Using the introduced categorical feature-based classification method, we identified the significance of the categorical features in classification accuracy. Also, SVM classification accuracy was increased from 65.8% to 84% by modifying the kernel width by using trial-And-error approach.