Skip to main content
. 2023 Jul 11;14:1162520. doi: 10.3389/fphys.2023.1162520

FIGURE 1.

FIGURE 1

Schematic outline of ML model development. (I) Building and calculation of a personalized electrophysiological ventricular model: (1) Processing of the CT imaging data. (2) Segmentation of the finite element meshes of the torso, lungs and ventricles; (2*) Personalization of the ventricular model: (A) Rule-based generation of myocardial fibers. (B) Assignment of the scar/fibrosis area in the ventricles (shown in back) and computing of the ventricular activation map at the baseline LBBB pattern and BiV pacing with clinical lead position. (3) Calculation of ECG signals from the ventricular activation map. (II) Development of a supervised machine learning classifier: creation of a dataset contacting combination of the clinical data and simulated features from the electrophysiological model from each of the 57 patients labeled into responders and non-responders, supervised training of a ML classifier and calculation of the ML-scores of CRT response.