Key result
Deep learning ECG model detects regional wall motion abnormalities with ~0.92 AUROC.
Why the study?
Regional wall motion abnormalities (WMA) reflect silent CAD even with preserved global ejection fraction, yet detection typically requires echocardiography or advanced imaging, limiting large-scale screening. Whether a deep learning model could detect WMA directly from standard 12-lead ECGs was evaluated.
Does a Vision Transformer deep learning model applied to standard 12-lead ECGs accurately detect regional wall motion abnormalities compared to echocardiography in patients undergoing paired ECG-echocardiography examinations?
Observational (n=43,010)
No
Does a Vision Transformer deep learning model applied to standard 12-lead ECGs accurately detect regional wall motion abnormalities compared to echocardiography in patients undergoing paired ECG-echocardiography examinations?
A Vision Transformer deep learning model applied to standard 12-lead ECGs detects regional wall motion abnormalities with high accuracy (AUROC 0.919) and a high negative predictive value (0.969), offering a potential scalable rule-out screening tool for silent coronary artery disease.
Answers are grounded in this paper and cite their sources.
Promising accuracy supports AI-assisted WMA detection; leaves open effects on outcomes and workflow integration.
Background Regional wall motion abnormalities (WMA) reflect myocardial ischemia or prior infarction and are associated with adverse cardiovascular outcomes. WMA may be present even in patients with preserved global ejection fraction, representing subclinical or silent coronary artery disease (CAD). Detection typically requires echocardiography or advanced imaging, limiting feasibility for large-scale screening. We evaluated whether a deep learning model could detect WMA directly from standard 12-lead electrocardiograms (ECGs). Methods We retrospectively analyzed paired ECG–echocardiography examinations collected between 2020 and 2023 at a tertiary academic medical center. WMA was defined as a wall motion score index >1.0 based on standard 17-segment echocardiographic analysis. ECG–echocardiography pairs were matched within ±14 days. A Vision Transformer architecture pretrained on unlabeled ECGs was fine-tuned to classify WMA using paired ECG–echo data. Data were split at the patient level into training, validation, and test sets (7:1:2). After matching and exclusions, 106,410 ECG–echocardiography pairs from 43,010 patients were included. The prevalence of WMA in the test cohort was 14.2%. Results The model achieved an AUROC of 0.919 (95% CI 0.910–0.928) in the test set. Using a threshold determined in the validation cohort where sensitivity equaled specificity, the model demonstrated sensitivity 0.838 (95% CI 0.815–0.858) and specificity 0.841 (95% CI 0.832–0.850). The positive predictive value was 0.466, while the negative predictive value reached 0.969, indicating strong potential as a rule-out screening tool. Conclusion A deep learning model using standard 12-lead ECGs demonstrated high accuracy for detecting regional wall motion abnormalities. ECG-based artificial intelligence may enable scalable, non-invasive screening for silent coronary artery disease and facilitate earlier identification of patients who may benefit from further cardiac evaluation.
No takes yet. Share an insight, caveat, or question.
Soh et al. (2026) conducted an observational in Regional wall motion abnormalities / silent coronary artery disease (n=43,010). Deep learning Vision Transformer model applied to 12-lead ECG vs. Standard 17-segment echocardiographic analysis (wall motion score index >1.0) was evaluated on Detection of regional wall motion abnormalities (wall motion score index >1.0) (AUROC 0.919, 95% CI 0.910-0.928). A deep learning 12-lead ECG model detected regional wall motion abnormalities with an AUROC of 0.919 (95% CI 0.910–0.928), sensitivity of 83.8%, and specificity of 84.1%.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: