Accuracy for Diagnosis of Heart Disease Using Sim Supervisor Algorithm

In this paper, a model for heart disease diagnosis using semi-supervised learning methods has been proposed. The proposed method, uses Semi-Supervised Support Vector Machine (S3VM) for heart disease diagnosis. Also, Expectation Maximization (EM) method is used for handling missing values in patient's observed data; which can improve diagnosis accuracy greatly. Thus, our proposed method for diagnosing heart disease, consists of three steps: "preprocessing by EM", "extracting features by PCA" and "disease detection using S3VM". In the evaluation of the proposed method, a real dataset containing medical information of 270 patients was used. Each data record had 13 attributes, which were reduced to 10 using the PCA algorithm. The results of the proposed model has been compared with other classification algorithms. Experimental results shows that using the proposed system, we could diagnose heart disease with an average accuracy of 83%. The results of this study show that the proposed method has better performance in terms of accuracy as well as in terms of specificity and sensitivity criteria and can be a useful tool for early heart disease diagnosis.