Use of Electrocardiograms to Identify Coronary Artery Disease: Cross-Validation of an Artificial Intelligence Model
Authors' conclusion
This study demonstrated the clinical utility of a deep learning artificial intelligence algorithm to analyze a digital 12-lead ECG to predict the presence of clinically significant CAD as determined by coronary angiography.
Study context
- Population
- Patients presenting for coronary angiography with a resting 12-lead digital electrocardiogram (ECG) recorded within 90 days prior to coronary angiography, 2019 to 2021patients with a resting 12-lead digital ECG within 90 days prior to coronary angiography
- Participants
- 16,476 patients
- Index test or model
- Artificial intelligence model applied to a resting digital 12-lead ECG vs coronary angiography (diameter stenosis ≥70% in LAD, LCx or RCA or ≥50% in left main)
- Target condition
- clinically significant CAD
Methodological details10-fold cross-validation, external validation set
- 10-fold cross-validation
- External validation set
Performance
As reported in the abstract
| Test or model | Area under the curve | Positive predictive value | Negative predictive value |
|---|---|---|---|
| Cross-validation cohort | 91.4% (95% CI: 89.4%-94.4%) | 91.7% (95% CI: 89.9%-93.4%) | 72.8% (95% CI: 69.6%-76.0%) |
| External validation | 92.4% (95% CI: 89.7%-95.1%) | 82.5% (95% CI: 75.9%-89.2%) | 88.1% (95% CI: 84.0%-92.1%) |
Area under the curve
Cross-validation cohort91.4% (95% CI: 89.4%-94.4%)
External validation92.4% (95% CI: 89.7%-95.1%)
Positive predictive value
Cross-validation cohort91.7% (95% CI: 89.9%-93.4%)
External validation82.5% (95% CI: 75.9%-89.2%)
Negative predictive value
Cross-validation cohort72.8% (95% CI: 69.6%-76.0%)
External validation88.1% (95% CI: 84.0%-92.1%)
Further findings
Prevalence of clinically significant coronary artery disease (CAD)
- Cross-validation cohort
- 64.5%
- External validation
- 36.0%
Original publication
Use of Electrocardiograms to Identify Coronary Artery Disease: Cross-Validation of an Artificial Intelligence Model. JACC Adv. Published 29 January 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →