An Intelligent Diagnosis System for Prediction of Heart Disease Risk based on Feature Selection and Ensemble Classification Techniques

Himanshu Sharma and Rohit Agarwal

In telemedicine and e-Healthcare application, intelligent disease diagnosis systems are adopted. Heart disease is most severe disease. The human death rate considerable increased by heart diseases. So, it requires detecting heart diseases more accurately. Many researchers are working on the same. Manual prediction is a complicated one. So using data mining technologies, an automated system can be developed. This helps to detect the heart diseases accurately as well as in less time consuming manner. There are various detection techniques available but their detection accuracy are not up to the mark. In this research work, intelligent Heart Disease prediction method is proposed to enhance the accuracy of the classifier. It is based on selection feature and ensemble classification methods. Expectation Conditional Maximization (IECM) is used remove the noisy information as well as to impute the missing attributes in the pre-processing stage. It preserves the attributes with more meaning. The number of features used to classify can be reduced using a feature selection process. Enhanced Kernel based Principal Component analysis (EKPCA) is used after the selection of features to enhance the accuracy of the classifier. At last, heart disease is predicted using an ensemble classifier and the single classifier is used to detect it. The experimental results shown the accuracy of the proposed method is high when compared with existing methods.

Volume 11 | 10-Special Issue

Pages: 666-675

DOI: 10.5373/JARDCS/V11SP10/20192856