Predicting Heart Diseases by Selective Machine Learning Algorithms

Abstract

Heart disease is among the leading causes of mortality worldwide. As a result, it’s critical to diagnose patients appropriately and promptly. Consequently, the objective of this paper was to predict heart diseases using selective machine learning algorithms.  The leverage technique was evaluated using the Cleveland heart disease dataset. In this study five classifiers were trained and tested with the unsmooth Cleveland dataset and the smooth Cleveland dataset. The results obtained showed all the classifiers performed better when tested with the smooth dataset with an accuracy of 98.11% than when tested with the unsmooth dataset with an accuracy of 89.71% The leverage technique performed better than works found in literature reviewed. These results show that feature engineering using data smoothing is effective for improved heart disease prediction.