Machine learning models for prediction of cardiovascular diseases

Abstract

Support Vector Machines (SVM) [9], Ada Boost (AB) [10], and Gradient Boosting (GB). Maximum Relevance, Minimum Redundancy (mRMR), Relief, and Least Absolute Shrinkage and Selection Operator (LASSO) are examples of fast correlation-based filters. The authors tested all attributes as well as the selected attributes generated by the above feature selection methods on the Cleveland heart disease dataset (CHDD) and the Hungarian heart disease dataset (HHDD). For their suggested framework, the authors were able to attain the greatest feasible model accuracy of 92.09 percent (all features) and 94.41 percent (selected features). The early detection of cardiac disease was demonstrated by several additional researchers.