Comparative Analysis of Supervised Machine Learning Algorithms for ECG Arrhythmia Detection using Small Dataset

Abstract

Automatic detection and analysis of arrhythmias from Electrocardiogram (ECG) signal is beginning to take a center stage in recent times, due to delay the nature of ECG signals, coupled with the subjective interpretation of these signals by cardiologist. Arrhythmia detection plays a vital role in diagnosing and managing cardiovascular diseases. With the advancements in machine learning techniques, various supervised algorithms have been employed to improve the accuracy of arrhythmia detection. However, each supervised ML algorithm has its strength and weakness in predicting the various classes of arrhythmia. This research study presents a comparative analysis of five popular supervised machine learning algorithms: support vector machine (SVM), K-Nearest Neighbors (KNN), Random Forest (RF), Naïve Baye (NB) and Decision Tree (DT) when applied to ECG arrhythmia detection with down sampled dataset. The goal is to evaluate and compare the performance of these algorithms in terms of accuracy, precision, recall, and F1 score. The study utilizes the MIT-BIH benchmark dataset, and experimental results provide insights into the strengths and limitations of each algorithm, aiding in the selection of the most suitable algorithm for accurate ECG arrhythmia detection. The Random Forest algorithm outperformed other algorithms in terms of accuracy, achieving an accuracy of 89.9% with RR interval based feature set.

Keywords

References