An Empirical Study of Heart Disease Prediction System Using Various Machine Learning Classification Algorithms

Abstract

In the world, cardiac disease is one of the leading causes  to death. Every year, 17.3 million people approximately die based on cardiac disease worldwide. Heart disease prediction system plays a significant role in health care utilization such as identifying individual expenses and disease risks for the patients. Early prediction of heart disease can prevent disease and can help lower the risk of heart diseases. Data mining (DM) techniques have a convincing role in the healthcare industry to empower health systems. This investigation inspects the early prediction of HD based on numerous DM techniques. Here, for experimentation, the dataset of the American Heart Association (AHA) is solicited to analyze the performance of this study and it was estimated with the six classifier evaluation metrics. The system was implemented using Weka tool and three different test cases were explored such as training set, Percentage Split (PS), and tenfold Cross-Validation (CV). The experiments verify that the random tree method could achieve a prediction accuracy of 100% with zero mean absolute error, which is greater than the other machine learning classification algorithms.