Article Impact Level: HIGH Data Quality: STRONG Summary of European Heart Journal, 44(8), 680–692. https://doi.org/10.1093/eurheartj/ehac617 Dr. Philippe Wouters et al.
Points
- A deep learning-based algorithm called FactorECG was developed to predict the outcome of cardiac resynchronization therapy (CRT) using electrocardiogram (ECG) features.
- FactorECG was trained on 1.1 million ECGs from 251,473 patients and outperformed current guideline ECG criteria and QRSAREA in predicting clinical outcomes.
- FactorECG identified several ECG features as significant predictors of poor outcome, including inferolateral T-wave inversion, smaller right precordial S- and T-wave amplitude, ventricular rate, and increased PR interval and P-wave duration.
- The study’s findings suggest that FactorECG has a superior discriminative ability for predicting clinical outcomes compared to current guideline ECG criteria and QRSAREA. Using an explainable algorithm allows for automated visualization of ECG features.
- An online visualization tool was created to provide interactive visualizations of FactorECG features, making it easier for clinicians to interpret and apply the findings of this study in their practice.
Summary
The study aimed to investigate the ability of a deep learning-based algorithm to predict the outcome of cardiac resynchronization therapy (CRT) using electrocardiogram (ECG) features. The algorithm, called FactorECG, was trained on 1.1 million ECGs from 251,473 patients and compressed the median beat ECG into 21 explainable factors. The pre-implantation ECGs of 1306 CRT patients were converted into their respective FactorECG, and the algorithm predicted the combined clinical endpoint of death, left ventricular assist device, or heart transplantation.
The results showed that FactorECG outperformed current guideline ECG criteria and QRSAREA in predicting clinical outcomes, with a c-statistic of 0.69 (95% CI 0.66-0.72) compared to 0.61 (95% CI 0.58-0.64) and 0.57 (95% CI 0.54-0.60), respectively. Adding 13 clinical variables did not significantly improve the FactorECG model’s performance. FactorECG identified several ECG features as significant predictors of poor outcome, including inferolateral T-wave inversion, smaller right precordial S- and T-wave amplitude, ventricular rate, and increased PR interval and P-wave duration.
The study’s findings suggest that FactorECG has a superior discriminative ability for predicting clinical outcomes compared to current guideline ECG criteria and QRSAREA. Using an explainable algorithm allows for automated visualization of ECG features, which could facilitate the rapid uptake of this personalized decision-making tool in CRT. The study highlights the potential of deep learning algorithms in the medical field, especially in predicting the outcomes of complex procedures such as CRT. The researchers created an online visualization tool to provide interactive visualizations of FactorECG features, making it easier for clinicians to interpret and apply the findings of this study in their practice.
Link to the article: https://academic.oup.com/eurheartj/article/44/8/680/6808667
References Wouters, P. C., Van De Leur, R. R., Vessies, M. B., Van Stipdonk, A. M. W., Ghossein, M. A., Hassink, R. J., Doevendans, P. A., Van Der Harst, P., Maass, A. H., Prinzen, F. W., Vernooy, K., Meine, M., & Van Es, R. (2023). Electrocardiogram-based deep learning improves outcome prediction following cardiac resynchronization therapy. European Heart Journal, 44(8), 680–692. https://doi.org/10.1093/eurheartj/ehac617
