| Model | Human mean-shape biventricular eikonal model coupled to CircAdapt. The geometry averages 100 healthy participants; the comparisons use one virtual geometry.PDF pp. 3, 4 | Reaction-eikonal electrophysiology in 24 anatomies with modelled proximal left bundle branch block (LBBB).PDF pp. 3, 4, 5, 6 | Human eikonal virtual population with a Purkinje network. Table 4 reports 70 ventricular geometries.PDF pp. 8, 9, 10, 12 | CMR-derived human biventricular digital twins with eikonal inference and monodomain simulation. Four control subjects and two subjects with hypertrophic cardiomyopathy.PDF pp. 2, 5, 8 |
|---|
| Comparator | Non-selective left bundle branch pacing (nsLBBP) versus right ventricular apical pacing (RVAP). The paper also models selective LBB pacing, septal, biventricular and His-bundle pacing.PDF pp. 3, 4 | Nominal non-selective left bundle branch area pacing (LBBAP) versus right ventricular apical pacing (RVA), with a separate sinus-initiated LBBB control and proximal/distal pacing positions.PDF pp. 3, 4, 5, 6 | LBBB without a conductive false tendon versus LBBB with a modelled direct false tendon to the ventricular free wall. Healthy geometry and alternative insertions are also tabulated.PDF pp. 8, 9, 10, 12 | Simulated versus recorded 12-lead ECG of the same subject. The paper also contrasts root-node, minimal Purkinje and full Purkinje configurations.PDF pp. 2, 5, 8 |
|---|
| Endpoint | Regional AT-95 (ms), the time to activate 95% of myocardial mass. Results are calculated separately for the left ventricular free wall, septum and right ventricular free wall.PDF pp. 3, 6 | BiV_TAT95 (ms), the node-weighted 95th percentile of activation times across both ventricles and the septum. The paper also reports LVAT95 for left ventricular nodes.PDF pp. 5, 6 | Simulated QRS duration (ms), defined as a myocardial activation-time surrogate.PDF pp. 10 | Pearson correlation of the simulated and clinical QRS waveforms. The results also describe QRS shortening with the full Purkinje configuration.PDF pp. 8, 10 |
|---|
| Time window | Electrical activation during each simulated pacing scenario, evaluated up to 95% of regional myocardial mass.PDF pp. 3, 6 | From stimulus time zero to the 95th percentile on the simulation clock. The first ventricular activation time is retained in that clock.PDF pp. 5, 6 | From the first myocardial activation to complete myocardial activation.PDF pp. 10 | Manually delineated QRS complex of the averaged clinical ECG (20 beats) and the corresponding simulated waveform.PDF pp. 8, 10 |
|---|
| Reported result | Left ventricular free-wall AT-95 is shorter during nsLBBP than during RVAP (Figure 5); the plotted values are used for direction only.PDF pp. 5, 6 | Nominal LBBAP versus RVA: BiV_TAT95 93.5 versus 96.5 ms and LVAT95 85.5 versus 98.5 ms (group medians, Table 3). The reported median paired reduction in BiV_TAT95 is 2.5 ms (Table 4).PDF pp. 10, 11 | The mean surrogate duration was 131.9 ± 9.8 ms for LBBB without a false tendon and 105.8 ± 7.1 ms with direct attachment to the ventricular free wall (Table 4).PDF pp. 12 | Most eikonal configurations report waveform correlation above 0.8. The full Purkinje configuration shortened simulated QRS in the monodomain comparisons (Figure 6).PDF pp. 9, 10 |
|---|
| Validation source | Clinical observations informed the simulation assumptions. The authors propose future in vivo comparison of the simulated measurements. The selected AT-95 contrast has no independent validation reported in the cited passages.PDF pp. 7, 9 | Within-study sensitivity analysis in three anatomies. The authors identify external patient-specific electrical validation as future work.PDF pp. 12, 13 | Previous literature supports the modelling approach. The authors report limited human evidence for direct validation of the false-tendon effect.PDF pp. 18 | Same-source comparison with clinical ECG. The target ECG is used during calibration. The selected results measure agreement with those data.PDF pp. 5, 8, 9 |
|---|