Comparison of Machine Learning Approaches for Classification of Cardiac Diseases

An accurate ECG (electrocardiogram) classification for diagnosis of various heart diseases poses a challenging problem for the researchers. This proposed work aims to classify various arrhythmia types based on analysis of ECG signals. Accurate as well as early detection of arrhythmia is pivotal in detection of heart disorders and aids in giving timely and appropriate treatment for the patient. The proposed work compares performance of various Machine Learning (ML) approaches for detection of various heart disorders. Decision Tree classifier was found to provide the results using Pan Tompkins algorithm giving an accuracy of 85.71% compared to 67.85% with Naive Bayes and 67.5% with kNN (k-Nearest Neighbor). Overall the Decision Tree classifier had an average accuracy of 25% higher than the kNN and Naïve Bayes.