Improving the performance of heart disease prediction system using ensembling techniques

Abstract

Machine learning is bringing a revolution in the healthcare domain. These algorithms have an immense capability to generate hidden insights from the data generated by the healthcare sector. These insights can be used to predict the risk of occurrence of fatal diseases in an early stage. Heart diseases pose a serious threat to the lives of people specifically in low- and middle-income group countries. Early detection of heart diseases using machine learning can be an effective way to prevent these diseases. In this research, a system was built to predict heart diseases. Freely available online dataset to predict heart diseases was used in the study. In this paper, a comprehensive effort has been made to enhance the performance of the prediction system. Ensembling techniques namely bagging and boosting have been applied. Experimental results prove that these techniques effectively improve the prediction accuracy of weak classifiers. The prediction system designed in this research shall prove to be a milestone in providing good quality affordable healthcare. This prediction system shall be made available in the cloud to ensure better accessibility.