A data-driven computational methodology for assessing ventricular ablation procedures

Ventricular tachycardia following myocardial infarction is often sustained by complex reentrant circuits that are challenging to characterize and treat using conventional electroanatomical mapping. Computational modeling provides a powerful complementary approach to understanding conduction pathway dynamics more effectively and supporting ablation strategies. Here, we present a reproducible and data-driven clinically guided computational framework for the retrospective analysis of post-infarction ventricular tachycardia and ablation procedures. The method integrates patient-specific electroanatomical mapping data-including local activation times, voltage maps, and electrograms-to build a personalized model that captures both structural and functional remodeling via a viability-based scalar field. A novel calibration procedure is introduced to locally estimate tissue conductivity, enabling accurate reproduction of observed activation patterns. The model is used to simulate arrhythmia inducibility and sustainability, and to retrospectively evaluate the impact of clinical radiofrequency ablation, accounting for lesion size and transmurality. In silico exploration of alternative ablation strategies is also performed to minimize lesion volume while maintaining arrhythmia suppression. The entire workflow is designed for rapid execution using a GPU-accelerated monodomain solver and is fully compatible with existing clinical practices, offering a practical tool for substrate interpretation and patient-specific ablation planning.