Support vector machines classification for discriminating coronary heart disease patients from non-coronary heart disease.

OBJECTIVE The present contribution concentrates on the application of support vector machines (SVM) for coronary heart disease and non-coronary heart disease classification. METHODS We conducted many experiments with support vector machine and different variables of low-density lipoprotein cholesterol (LDLC), high-density lipoprotein cholesterol (HDLC), total cholesterol (TC), triglycerides (TG), glucose and age (dataset 346 patients with completed diagnostic procedures). Linear and non-linear classifiers were compared: linear discriminant analysis (LDA) and SVM with a radial basis function (RBF) kernel as a non-linear technique. RESULTS The prediction accuracy of training and test sets of SVM were 96.86% and 78.18% respectively, while the prediction accuracy of training and test sets of LDA were 90.57% and 72.73% respectively. The cross-validated prediction accuracy of SVM and LDA were 92.67% and 85.4%. CONCLUSION Support vector machine can be used as a valid way for assisting diagnosis of coronary heart disease.