Learning to Predict Global Atrial Fibrillation Dynamics from Sparse Measurements

2025-02-13

AI Summary

Clinical catheter ablation for atrial fibrillation (AF) is hindered by low spatial resolution and limited coverage of contact mapping, impeding comprehensive characterization of global electrophysiological dynamics. Method: We propose FibMap, a graph recurrent neural network that reconstructs whole-atrial AF dynamics with high fidelity from sparse electrophysiological signals—requiring only 10% atrial surface coverage. FibMap jointly models spatial topology via graph neural networks, captures temporal dynamics with RNNs, and incorporates sparse-signal priors and biophysically grounded constraints to yield an interpretable, mechanistic state-space representation. Results: Evaluated on 51 non-contact mapping datasets, FibMap reduces mean absolute error by 210% and improves phase singularity tracking accuracy by an order of magnitude. Crucially, when applied to real contact mapping data, it achieves reconstruction fidelity comparable to non-contact mapping—establishing a novel paradigm for patient-specific, precision ablation.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Catheter ablation of Atrial Fibrillation (AF) consists of a one-size-fits-all treatment with limited success in persistent AF. This may be due to our inability to map the dynamics of AF with the limited resolution and coverage provided by sequential contact mapping catheters, preventing effective patient phenotyping for personalised, targeted ablation. Here we introduce FibMap, a graph recurrent neural network model that reconstructs global AF dynamics from sparse measurements. Trained and validated on 51 non-contact whole atria recordings, FibMap reconstructs whole atria dynamics from 10% surface coverage, achieving a 210% lower mean absolute error and an order of magnitude higher performance in tracking phase singularities compared to baseline methods. Clinical utility of FibMap is demonstrated on real-world contact mapping recordings, achieving reconstruction fidelity comparable to non-contact mapping. FibMap's state-spaces and patient-specific parameters offer insights for electrophenotyping AF. Integrating FibMap into clinical practice could enable personalised AF care and improve outcomes.

Problem

• Reconstruct global AF dynamics
• Improve atrial fibrillation mapping
• Enable personalized AF treatment

Innovation

• Graph recurrent neural network
• Reconstructs global AF dynamics
• Personalised electrophenotyping insights