Can in silico models predict drug-induced cardiac risk in vulnerable populations?

Figures

• 

  Fig. 1. Overview of study methodology. The baseline population captures anatomical variability, including male and female anatomies, as well as specific cardiac pathologies: heart failure (HF), ischaemia (ISC), myocardial infarction (MI), dilated cardiomyopathy (DCM), and hypertrophic cardiomyopathy (HCM). For each combination of anatomy and physiological condition, 32 virtual patients are generated by varying ionic conductances. Each virtual patient is subsequently evaluated under baseline and drug-exposure (benchmark and contraindicated conditions).

• 

  Fig. 2. Reconstructed hearts from the Visible Heart Lab library, including AG hearts. Hearts H02 and H05 were both used for healthy, HF, ISC and MI conditions. ∗Scars were included just for ISC and MI conditions.

• 

  Fig. 3. Ruled-based fibres short axis slice of a HCM heart. On the left, original fibre distribution. On the right, 50% fibre misalignment.

• 

  Fig. 4. APs of the same single representative virtual subject from different populations under baseline conditions. For diseases with localized manifestations (ISC and MI), the APs correspond to the affected tissue region.

• 

  (a). Female pseudo-ECG.

• 

  (b). Male pseudo-ECG.

• 

  (a). Healthy populations.

• 

  (b). Pathological populations.

• 

  (a). Average regression models.

• 

  (b). Sex-specific regression models.

• 

  Fig. 8. Computational pseudo-ECGs (leads I, II, and III) illustrating different arrhythmic events. Each trace begins with three stabilization beats under baseline conditions, followed—after a non-simulated transition period (indicated by a break)—by the ECG trace under drug exposure. This highlights the progression from normal rhythm to arrhythmia onset. Asystole trace shows very low amplitude, understood as residual noise below the threshold for effective ventricular activation.

• 

  (a). Healthy populations.