Validation of a Digital Twin Model for Three‐Dimensional Noninvasive Functional Substrate Prediction in Scar Related Ventricular Tachycardia

Background High density electroanatomical mapping (EAM) is the standard for identifying functional substrate in scar‐related ventricular tachycardia (VT), but it is limited by its two‐dimensional nature and increased procedural complexity. Imaging‐based strategies provide anatomical characterization but lack functional information. We evaluated a novel digital twin model (CardioMat) to non‐invasively predict functional substrate in three‐dimensions. Objective To validate a novel digital twin model for noninvasive prediction of functional substrate in scar‐related ventricular tachycardia. Methods Fifteen patients with ischemic or nonischemic cardiomyopathy undergoing VT ablation were retrospectively analyzed. High‐density EAM, cardiac magnetic resonance (CMR) post‐processing with ADAS3D, and CardioMat modeling were performed. CardioMat‐generated three‐dimensional activation maps were integrated into the CARTO EAM suite. Deceleration zones (DZs) identified by CardioMat were compared with EAM‐derived DZs and CMR‐based heterogeneous tissue channels (HTCs) and scar areas. Results A total of 255 cardiac segments were analyzed. CardioMat‐derived DZs demonstrated high diagnostic accuracy (79% sensitivity, 93% specificity, 81% PPV, 92% NPV, 89% accuracy) for predicting EAM‐DZ localization, outperforming CMR scar and HTCs. Multivariate analysis confirmed CardioMat DZs are the strongest independent predictor of EAM‐DZ localization (OR 17.4, 95% CI 6.9–43.8, p > 0.001). Conclusions This study provides preliminary evidence supporting the feasibility of a digital twin model for noninvasive functional substrate characterization. CardioMat integrates anatomical and functional information, outperforming conventional CMR‐based assessment. Prospective multicenter validation is needed to establish its role in VT ablation guidance.