Machine Learning Based Categorization of Heart Sounds for Cardiac Arrhythmia Detection

Early detection of cardiac arrhythmias is vital for preventing severe cardiovascular conditions and improving patient outcomes. This research focusses on the implementation of machine learning models for classifying heart sounds (HS) to identify arrhythmias effectively. Heart sound signals were analyzed and classified using five machine learning models: XGBoost, Support Vector Machine (SVM), Extra Trees, LightGBM, and Logistic Regression. These models were trained and estimated using benchmarks such as accuracy, precision, recall, and F1-score to judge their performance in discriminating rhythmic and arrhythmic heart sounds. The results indicate that Extra Trees achieved the highest accuracy, followed by LightGBM and Logistic Regression, while XGBoost and SVM demonstrated comparatively lower accuracies.