Using Machine Learning to Predict Adverse Events in Patients Undergoing Transcatheter Left Atrial Appendage Occlusion

doi:10.1016/j.jacadv.2025.102540

Arrhythmias & electrophysiologyDiagnostic or prognostic study

Using Machine Learning to Predict Adverse Events in Patients Undergoing Transcatheter Left Atrial Appendage Occlusion

JACC AdvPublished 20 Jan 2026

Registry-based prediction model development and validation (logistic regression, least absolute shrinkage and selection operator (LASSO), XGBoost)
Randomly selected 70% development and 30% validation cohorts
16 variables from the previous Left Atrial Appendage Occlusion (LAAO) risk model and an expanded set of 51 variables

Read the original article ↗

Authors' conclusion

In a nationwide LAAO cohort, XGBoost enhanced discrimination of composite MAE and several individual events. Prediction of rare mortality events was improved, although this was not consistently the case for other rare outcomes.

Study context

Population: Patients undergoing transcatheter LAAO in the National Cardiovascular Data Registry Left Atrial Appendage Occlusion (LAAO) Registry
Participants: 81,703 LAAO procedures

Comparison: XGBoost vs logistic regression (LR) and LASSO for predicting combined in-hospital major adverse events (MAEs) and individual events

Performance

As reported in the abstract

Composite MAE rate · in-hospital

Composite MAE rate

1.39%

Individual event rate

0.02% to 1.13%

Further findings

AUC

Outcome	LR	LASSO	XGBoost
Discrimination for MAE, original model variables (validation AUC) · XGBoost performed best	0.630 (95% CI 0.608–0.642)	0.638 (95% CI 0.626–0.670)	0.648 (95% CI 0.626–0.670)
Discrimination for MAE, expanded variables (AUC)	0.515 (95% CI 0.501–0.529)	0.644 (95% CI 0.628–0.660)	0.653 (95% CI 0.635–0.671)

Discrimination for MAE, expanded variables (AUC): XGBoost performed marginally better than LASSO; LR performed poorly

Individual and infrequent events

Performance using all methods declined for infrequent events
XGBoost generally outperformed other methods for individual events, particularly with expanded variables, though not necessarily for rare events
Mortality prediction using XGBoost was incrementally better

Original publication

Using Machine Learning to Predict Adverse Events in Patients Undergoing Transcatheter Left Atrial Appendage Occlusion. JACC Adv. Published 20 January 2026. doi:10.1016/j.jacadv.2025.102540

Read the original article ↗PubMedCite RIS · BibTeX

Summarised from the published abstract; every number is checked against it