Rapid prediction of cardiac activation in the left ventricle with geometric deep learning: a step towards cardiac resynchronization therapy planning

1

Cardiac resynchronization therapy (CRT) is effective for heart failure, but one-third of patients do not respond due to lead placement issues.
2

Two geometric deep learning models, GNN and GINO, were developed to predict left ventricular activation time maps in real time.
3

The GINO model outperformed the GNN model on synthetic cases, achieving a 1.38% error compared to 2.44%.
4

Both models effectively identified optimal pacing sites and recovered subject-specific parameters from noisy data.
5

An interactive web-based interface was created to support personalized pre-procedural optimization for CRT.

Npj Digital Medicine

by Ehsan Naghavi, Haifeng Wang, Vahid Ziaei-Rad, Julius Guccione, Ghassan Kassab, Vishnu Boddeti, Seungik Baek, Lik-Chuan Lee
February 7, 2026

1