Arrhythmias & electrophysiology
Electrocardiogram (ECG)-only explainable deep learning algorithm predicts the risk for malignant ventricular arrhythmia in phospholamban cardiomyopathy
Authors' conclusion
Our deep learning-based algorithm based on ECG data only accurately predicts the occurrence of MVA in PLN p.(Arg14del) carriers, enabling more efficient stratification of patients who need additional diagnostic testing and follow-up.
Study context
- Population
- Phospholamban (PLN) p.(Arg14del) variant carriers without malignant ventricular arrhythmia (MVA) at baseline
- Participants
- 679 participants
- Comparison
- ECG-only deep learning model vs current multimodality prediction model vs conventional ECG parameter model
- Outcome
- Prediction of malignant ventricular arrhythmia (MVA)
Methodological detailsexplainable deep learning, variational auto-encoder trained on 1.1 million ECGs, Cox regression
- Explainable deep learning
- Variational auto-encoder trained on 1.1 million ECGs
- Cox regression
Performance
As reported in the abstract
MVA prediction (C statistic)
FactorECG model
0.79
C statistic · 95% CI 0.76–0.83
Current prediction model
0.83
C statistic · 95% CI 0.79–0.88
Conventional ECG model
0.65
C statistic · 95% CI 0.58–0.73
FactorECG model vs Current prediction model: P = 0.054 · FactorECG model vs Conventional ECG model: P < 0.001
Further findings
Clinical simulation: 2-step approach (ECG-only screening, then full workup)
- Less additional diagnostics
- 60%
- Outperformed the multimodal prediction model in all patients
Original publication
ECG-only explainable deep learning algorithm predicts the risk for malignant ventricular arrhythmia in phospholamban cardiomyopathy. Heart Rhythm. Published 23 February 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →