Heart failure & cardiomyopathies
A Novel ECG-Based Deep Learning Algorithm to Predict Cardiomyopathy in Patients With Premature Ventricular Complexes
Authors' conclusion
Deep-learning on the 12-lead ECG alone can accurately predict new-onset cardiomyopathy in patients with PVCs independent of PVC burden. Model prediction performed well across sex and race, relying on the QRS complex/ST-segment in sinus rhythm, not PVC morphology.
Study context
- Population
- Adults with documented premature ventricular complexes (PVCs)of 383,514 electrocardiogram (ECGs)
- Participants
- 14,241 adults
- Outcome
- First diagnosis of LVEF ≤40%
- Follow-up
- Within 6 months
Methodological detailsmulticentre (5 hospitals), internal training/testing at one hospital, external validation at the others
- Multicentre (5 hospitals)
- Internal training/testing at one hospital, external validation at the others
Main finding
As reported in the abstract
Prediction of LVEF reduction to ≤40% within 6 months
Deep-learning model AUC for left ventricular ejection fraction (LVEF) ≤40%
0.79
95% CI 0.77–0.81
Incidence of LVEF reduction to ≤40% within 6 months
22.9%
Observational data: associations, not effects.
Further findings
Explainability (Grad-CAM)
- Highlighted the sinus rhythm QRS complex-ST segment
Resolution of cardiomyopathy after successful PVC ablation
- Post-ablation LVEF improvement with resolution of cardiomyopathy
- 89%
Original publication
A Novel ECG-Based Deep Learning Algorithm to Predict Cardiomyopathy in Patients With Premature Ventricular Complexes. JACC Clin Electrophysiol. Published 19 July 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →