Predicting the When : Multimodal AI for Time-to-Recurrence Analysis After Atrial Fibrillation Ablation.

Area of Science:

• Cardiology
• Artificial Intelligence
• Medical Imaging

Background:

• Catheter ablation is a primary treatment for atrial fibrillation (AF), but recurrence rates remain high.
• Current management strategies lack tools to predict individual patient recurrence timing.
• Multimodal artificial intelligence (AI) offers a novel approach to address this clinical need.

Purpose of the Study:

• Develop a predictive model for time-to-AF-recurrence post-ablation.
• Utilize pre-procedural bi-atrial imaging, clinical data, and procedural characteristics.
• Employ a multimodal AI and survival analysis framework for accurate predictions.

Main Methods:

• Retrospective analysis of 437 AF patients undergoing catheter ablation.
• Training the MARTA-AF model on pre-procedural bi-atrial images and patient data.
• Integrating the model into a survival framework for time-varying recurrence risk estimation.

Main Results:

• MARTA-AF accurately predicted time-varying AF recurrence risk up to three years post-ablation.
• Patients were effectively stratified into low- and high-risk groups with sustained discrimination.
• Right atrial shape features improved prediction accuracy, and key predictors were identified.

Conclusions:

• MARTA-AF provides individualized, time-varying AF recurrence risk forecasts.
• The model enables stratification into clinically meaningful risk groups for proactive management.
• This AI framework can transform post-ablation care and inform pre-ablation clinical decisions.