Year: 2025
Type: article
Abstract: Background and Objectives: Mitral stenosis (MS) increases the risk of atrial fibrillation (AF) and thromboembolic events, requiring appropriate anticoagulation.However, echocardiography, essential for MS diagnosis, is not always immediately available.This study evaluated the performance of an artificial intelligence (AI)-enabled model analyzing printed ECG images to predict moderate to severe MS.Methods: A deep learning-based algorithm was developed using approximately 250,000 ECGs from 140,000 patients at Hospital A, paired with echocardiographic data.An external test dataset was retrospectively collected from Hospital B from 2012 to 2024, including patients (age ≥18) who underwent ECG and echocardiography within one year.Patients with moderate or severe MS were included, and 1:3 age-, sex-, and AFmatched controls were randomly selected.AI-predicted MS scores were compared to echocardiographic findings.Results: The internal dataset included 1,936 ECGs from 610 patients with moderate to severe MS, in which the model achieved an AUC of 0.965.For external validation, 123 patients with MS and 369 matched controls were analyzed.The MS group had a mean age of 64 years, 22% male, and 56% with atrial fibrillation; none differed significantly from the control group.The AI model achieved an AUC of 0.822 for detecting overall MS, and 0.847 for severe MS specifically.When clinical parameters including age, sex, AF status, and underlying comorbidities were integrated, the AUC improved to 0.974.AI-ECG scores increased with MS severity, showing a significant difference between moderate and severe MS (p 5 0.021).Subgroup analysis showed no significant performance differences by rhythm or sex, but the model performed significantly better in patients aged ,60 years than ≥60 years (AUC 0.886 vs. 0.776, p 5 0.014).Conclusion: AI-ECG based prediction of MS is feasible and demonstrates promising accuracy.This approach may support early detection and guide appropriate anticoagulation decisions when echocardiography is not readily available.
Source: Heart Rhythm O2
Authors: Shekhar Singh, Younghoon Kwon, Ibrahim Karabayir, Liam Butler, Ali Azarbarzin, Arti Taneja, Susan Redline, Susan Heckbert, Bruce Psaty, Jennifer Lobo, Kenneth C. Bilchick, Pamela Lutsey, Lin Yee Chen, Elsayed Z. Soliman, Oguz Akbilgic
Language: English
Topic: ECG Monitoring and Analysis
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: gold
Estimated APC paid: $2,150
OpenAlex ID: W4413963462