Why the study?
Left atrial thrombus or spontaneous echo contrast are major contributors to cardiogenic embolism in NVAF, prompting the development and validation of an interpretable machine learning risk prediction model.
Does a machine learning-based logistic regression model improve the prediction of left atrial thrombus or spontaneous echo contrast compared to the CHA2DS2-VASc score in patients with non-valvular atrial fibrillation?
Population
1,222 NVAF patients scheduled for catheter ablation
Comparison
Optimal machine learning model vs CHA2DS2-VASc scoring system
Design
Retrospective single-center cohort study
Key result
A machine learning-based logistic regression model significantly outperformed the CHA2DS2-VASc scoring system in predicting left atrial thrombus or spontaneous echo contrast in NVAF patients (AUC 0.831 vs 0.650, P < 0.001).
Authors
Loading...
May enhance LAT/SEC prediction in NVAF; hypothesis-generating and should not yet change practice.
Observational (n=1,078)
No
Does a machine learning-based logistic regression model improve the prediction of left atrial thrombus or spontaneous echo contrast compared to the CHA2DS2-VASc score in patients with non-valvular atrial fibrillation?
Absolute Event Rate: 0.831% vs 0.65%
p-value: p=<0.001
A machine learning-based logistic regression model incorporating six clinical and echocardiographic features significantly outperforms the CHA2DS2-VASc score in predicting left atrial thrombus or spontaneous echo contrast in patients with non-valvular atrial fibrillation.
Huang et al. (2025) conducted an observational in Non-valvular atrial fibrillation (NVAF) (n=1,078). Machine learning (logistic regression) model vs. CHA2DS2-VASc scoring system was evaluated on Area under the receiver operating characteristic curve (AUC) for predicting LAT/SEC (95% CI 0.790-0.868, p=<0.001). A machine learning-based logistic regression model significantly outperformed the CHA2DS2-VASc scoring system in predicting left atrial thrombus or spontaneous echo contrast in NVAF patients (AUC 0.831 vs 0.650, P < 0.001).