Understanding the Role of Fibrosis in Atrial Arrhythmogenesis: Towards Biomarker Discovery through Personalised Computational Models

Atrial fibrillation (AF) is the most common cardiac arrhythmia and a major contributor to healthcare costs. Its initiation and persistence are strongly influenced by atrial fibrosis, yet accurately identifying patient-specific fibrotic patterns remains a major clinical and computational challenge. Current imaging and electro-anatomical mapping techniques provide only limited and indirect information about the fibrotic substrate, thereby hindering the development of personalized treatment strategies. 

This project aims to develop data-driven methodologies to infer realistic, patient-specific atrial fibrosis distributions from clinical electrograms and local activation maps, and to investigate their impact on cardiac electrophysiology. We will combine cardiac digital twins with high-performance computing and machine learning, generating large virtual cohorts of anatomically accurate atrial models with diverse fibrotic patterns and corresponding simulated clinical signals. These datasets will be used to train deep-learning models that learn the relationship between electrical measurements and underlying fibrotic structure, enabling fibrosis reconstruction across different anatomies.

The resulting personalized digital twins will be calibrated and validated against clinical data and used to study mechanisms linking fibrosis to electrical remodeling, identify novel AF biomarkers, and explore simulation-assisted optimization of catheter ablation strategies. By tightly integrating clinical data, physics-based modeling, and machine learning, this research will advance digital-twin based cardiology and support precision treatment planning for AF.

 

Project funding

Funding EUR 490.000,-