Machine Learning Techniques Used for Diagnosing Cardiac Abnormalities using Electrocardiogram

Abstract

Atrial fibrillation (AF) damages around 1–2% of the human body and is the most serious heart arrhythmia in human healthcare. As per the guidelines of World Health Organization (WHO), coronary thrombosis (CT) is the main cause of mortality throughout world and in India. As a consequence of coronary artery disorder, high blood pressure, alcoholic abuse, and a life in emotional stress, heart arrhythmias, or irregular heartbeats, are among the most prevalent CTs. In addition to the CTs described, an irregularity in heart rhythm is produced by mental stress in long duration, which is a difficult problem for researchers to solve. One of the most important areas of study after the development of the electrocardiogram (ECG) and robust machine learning algorithms is the early detection of cardiac arrhythmia using automated electronic methods. As a gold standard for studying heart function, cardiologists and researchers rely on the ECG because it records changes in electrical activity connected to the cardiac cycles. The research was conducted on ECG analysis and categorization utilizing classic and novel artificial intelligence (AI) methods. As a result of the research, a detailed report is expected. Recent years have seen a rise in the use of AI methods to detect arrhythmia signs automatically and early. This chapter examines the literature of the past few years to assess the performance of artificial intelligence and other computer-based systems for processing ECG data for the detection of heart problems. Performance measurements such as specificity, sensitivity, accuracy, positive predictive value (PPV), etc. are used to evaluate machine learning (ML) and deep learning (DL) methods for detecting the AF from ECG signal analysis and categorization. These are crucial in the early detection of AF because accurate detection of heart problems can reduce mortality. © 2025 Apple Academic Press, Inc.