Volume 30, Issue 169, March 2026

AI-driven interpretation of electrocardiograms in the diagnosis of common cardiovascular disorders: a systematic review

Anna Mazur^1♦, Julia Brodziak¹, Wiktoria Ciszewska¹, Michał Dworak¹, Katarzyna Fojcik¹, Zofia Kosztyła-Czech¹, Marta Kowalska¹, Mateusz Matyja², Wiktor Werenkowicz¹, Tomasz Wiśniewski¹

¹Medical University of Silesia, Poniatowskiego 15, 40-055 Katowice, Poland
²Wojewodzki Szpital Specjalistyczny nr 5 w Sosnowcu, Plac Medykow 1, 41-200 Sosnowiec, Poland

^♦Corresponding author
Anna Mazur, Medical University of Silesia, Poniatowskiego 15, 40-055 Katowice, Poland

ABSTRACT

Electrocardiography (ECG) is widely used in everyday cardiology practice, but its interpretation can be difficult, especially when electrical abnormalities are mild or atypical. Artificial intelligence (AI) has become an important tool in ECG interpretation over the past 10 years. These methods can improve the detection of arrhythmias, ischemia, heart failure, and structural heart diseases. In this article, we review studies that analyzed how AI can support ECG-based diagnosis across a range of common cardiovascular (CV) conditions. We identified 26 articles that met the inclusion criteria. Most of the included studies relied on deep learning approaches. Their performance was encouraging, especially in detecting atrial fibrillation (AF), myocardial infarction (MI), left ventricular dysfunction, and various cardiomyopathies. New research also supports the use of artificial intelligence-assisted ECGs for detecting pulmonary hypertension, treatment-related cardiotoxicity, cardiovascular risk estimation, and predicting diastolic dysfunction. A major obstacle is that scientists trained many models on retrospective datasets from a single center. Only a few of them underwent robust external validation. Despite this, AI-ECGs may help clinicians in everyday work. In the future, research should involve larger, more diverse patient groups, standardized reporting standards, and prospective validation studies. These steps are necessary to understand whether AI-ECG tools can improve clinical outcomes.

Keywords: artificial intelligence; electrocardiogram; deep learning; cardiovascular diseases; arrhythmia

Medical Science, 2026, 30, e53ms3794

PDF

DOI: https://doi.org/10.54905/disssi.v30i169.e53ms3794

Published: 12 March 2026