Ensemble Learning for Coronary Heart Disease Prediction

Abstract

Disease diagnosis is an important and tedious job in medicine. Nowadays, cardiovascular diseases are the major cause of death in the world, especially in the low income countries. The variation in blood pressure, blood sugar and heart rate leads to heart disease that in turn leads to narrowing of blood vessels. Therefore, diagnosing of heart disease has foreseen massive amount of attention globally. There is a huge quantity of data in healthcare sector; therefore the need is to convert this raw data into useful information. Although medical practitioners have been using the conventional clinical methods such as blood tests and electrocardiography for heart disease prediction, however computerized diagnosis systems are also in use. Some currently existing studies have applied various machine learning techniques for heart disease prediction purpose. But the performance achieved by most of the basic machine learning algorithms is not up to the mark. There are different types of heart diseases and one of them is the coronary heart disease or CHD. In this paper various ensemble machine learning classifiers have been used for coronary heart disease prediction. Ensemble classifiers combine varying set of individual models to improve the reliability and stability of individual algorithms. Ensemble classifiers combine the output from multiple classifiers to get the aggregated output with extended prediction accuracy. Ensemble classifiers that have been used in this study include adaboost, majority voting, weighted average, bagging and gradient boosting. According to the results the classifier that came up with the highest prediction accuracy was bagging ensemble classifier predicting with an accuracy of 85% on the given dataset.