A comprehensive review of AI-Based detection of Arrhythmia using Electrocardiogram (ECG)

Abstract

The success of AI-assisted decision-making systems over traditional methods has driven extensive research across various real-world applications. In the past decade, the application of AI systems for analysing physiological signals, particularly electrocardiograms (ECG), has attracted considerable attention. While several survey papers have explored this domain, they often face limitations, for instance outdated research coverage, narrow scope, inadequate evaluation of study quality and publication credibility or a lack of statistical insights. To address these gaps, this review rigorously selected research articles from high-impact journals and top-tier conferences, ensuring reliable and validated findings. We comprehensively reviewed 219 research articles, making this paper a valuable resource for researchers interested in the intersection of AI and ECG analysis. Our review provides an in-depth analysis of employed techniques, obtained results, and emerging trends, offering insights beneficial to researchers at all levels. Additionally, we present a statistical analysis of the reviewed studies to offer a broader understanding of this research area. A key contribution of this paper is the application of Pearson’s correlation to examine relationships among performance metrics such as accuracy, sensitivity, specificity, and F1-score. This analysis highlights how these metrics interact and influence each other across various methodologies, offering deeper insights into model performance and optimisation strategies in ECG analysis. Finally, we address existing challenges and propose new research directions for further exploration.