Arrhythmias & electrophysiology
Expert-Level Automated Diagnosis of the Pediatric ECG Using a Deep Neural Network
Authors' conclusion
Our model provides expert-level automated diagnosis of the pediatric 12-lead ECG, which may improve access to care.
Study context
- Population
- All patients with an ECG at Boston Children's Hospital read by an experienced pediatric cardiologist (≥5,000 reads) between 2000 and 2022patients in the main cohort, 583,134 electrocardiograms (ECGs)
- Participants
- 201,620 patients
- Exposure
- AI-ECG model vs commercial software interpretations
- Outcome
- Composite of any ECG abnormality (detecting normal vs abnormal ECG)
Main finding
As reported in the abstract
| Group | AUROC | AUPRC |
|---|---|---|
| AI-ECG model | 0.94 | 0.96 |
AI-ECG model
AUROC0.94
AUPRC0.96
Observational data: associations, not effects.
Secondary outcomes
| Outcome | AI-ECG model | Commercial software interpretations | |
|---|---|---|---|
| Wolff-Parkinson-White syndrome (WPW)Secondary outcome | |||
| AUROC | 0.99 | Not reported in the abstract | commercial software interpretations: not reported in the abstract |
| AUPRC | 0.88 | Not reported in the abstract | commercial software interpretations: not reported in the abstract |
| Prolonged QTcSecondary outcome | |||
| AUROC | 0.96 | Not reported in the abstract | commercial software interpretations: not reported in the abstract |
| AUPRC | 0.63 | Not reported in the abstract | commercial software interpretations: not reported in the abstract |
Readjudication of ECGs with AI-ECG/original cardiologist read discordance: blinded expert readers more likely to agree with AI-ECG classification than the original reader
- Any abnormality
- P = 0.001
- WPW
- P = 0.01
- Prolonged QTc
- P = 0.07
Cohort characteristics
- Male
- 49%
- Known congenital heart disease
- 11%
- Median age
- 11.7 years [Q1-Q3: 3.1-16.9 years]
- Any ECG abnormality
- 56%
- WPW
- 1.0%
- Prolonged QTc
- 5.3%
Original publication
Expert-Level Automated Diagnosis of the Pediatric ECG Using a Deep Neural Network. JACC Clin Electrophysiol. Published 17 March 2025. doi:10.
Summarised from the published abstract; every number is checked against it