Fine-tuning an ECG Foundation Model to Predict Coronary CT Angiography Outcomes

2025-11-29

AI Summary

This study addresses the clinical limitations of coronary computed tomography angiography (CCTA)—including ionizing radiation exposure, high equipment requirements, and stringent patient cooperation—by proposing an interpretable AI-powered electrocardiogram (ECG) model for noninvasive prediction of severe or occlusive stenosis in four major coronary arteries: right coronary artery (RCA), left main (LM), left anterior descending (LAD), and left circumflex (LCX). Methodologically, the model leverages fine-tuning of a pre-trained ECG foundation model and integrates gradient-weighted class activation mapping (Grad-CAM) to elucidate electrophysiological associations between ECG waveform features and specific coronary lesions. In multicenter internal and external validation cohorts, the model achieves AUCs up to 0.971 across the four vessels. Notably, it maintains robust performance on normal-appearing ECGs and demonstrates strong risk stratification capability, underscoring its clinical utility and generalizability.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Coronary artery disease (CAD) remains a major global health burden. Accurate identification of the culprit vessel and assessment of stenosis severity are essential for guiding individualized therapy. Although coronary CT angiography (CCTA) is the first-line non-invasive modality for CAD diagnosis, its dependence on high-end equipment, radiation exposure, and strict patient cooperation limits large-scale use. With advances in artificial intelligence (AI) and the widespread availability of electrocardiography (ECG), AI-ECG offers a promising alternative for CAD screening. In this study, we developed an interpretable AI-ECG model to predict severe or complete stenosis of the four major coronary arteries on CCTA. On the internal validation set, the model's AUCs for the right coronary artery (RCA), left main coronary artery (LM), left anterior descending artery (LAD), and left circumflex artery (LCX) were 0.794, 0.818, 0.744, and 0.755, respectively; on the external validation set, the AUCs reached 0.749, 0.971, 0.667, and 0.727, respectively. Performance remained stable in a clinically normal-ECG subset, indicating robustness beyond overt ECG abnormalities. Subgroup analyses across demographic and acquisition-time strata further confirmed model stability. Risk stratification based on vessel-specific incidence thresholds showed consistent separation on calibration and cumulative event curves. Interpretability analyses revealed distinct waveform differences between high- and low-risk groups, highlighting key electrophysiological regions contributing to model decisions and offering new insights into the ECG correlates of coronary stenosis.

Problem

• Predicts severe coronary artery stenosis using AI-ECG as an alternative to CT angiography.
• Identifies culprit vessels and stenosis severity from ECG data for CAD diagnosis.
• Provides interpretable AI model to reduce reliance on invasive imaging methods.

Innovation

• Fine-tuned ECG foundation model for CCTA outcome prediction
• Interpretable AI-ECG model predicts severe coronary stenosis
• Model uses ECG waveforms to identify key stenosis regions