Quantifying variabilities in cardiac digital twin models of the electrocardiogram

TL;DR

This study investigates how residual beat-to-beat variability, cardiac anatomy, torso conductivities, and electrode placements affect ECG morphology in cardiac digital twin calibrations. Using a subject-specific, biophysically detailed heart–torso forward model calibrated to measured ECGs, the authors systematically vary heart position, orientation, size, torso heterogeneity, and lead placements to quantify impact on 12-lead ECG features. They find that diagnostically relevant morphology is robust to these uncertainties, with residual beat-to-beat variability producing a narrow envelope and certain leads showing sensitivity mainly to heart-to-lead distance or activation patterns. The findings support the feasibility of accurate, ensemble-aware CDT calibration and provide guidance on which factors most influence ECG morphology, while acknowledging limitations such as single-subject anatomy and potential non-uniqueness in parameter identification. Overall, the work suggests robust ECG-based calibration of CDTs under realistic uncertainties, informing clinical translation and design of calibration workflows.

Abstract

Cardiac digital twins (CDTs) of human cardiac electrophysiology (EP) are digital replicas of patient hearts that match like-for-like clinical observations. The electrocardiogram (ECG), as the most prevalent non-invasive observation of cardiac electrophysiology, is considered an ideal target for CDT calibration. Recent advanced CDT calibration methods have demonstrated their ability to minimize discrepancies between simulated and measured ECG signals, effectively replicating all key morphological features relevant to diagnostics. However, due to the inherent nature of clinical data acquisition and CDT model generation pipelines, discrepancies inevitably arise between the real physical electrophysiology in a patient and the simulated virtual electrophysiology in a CDT. In this study, we aim to qualitatively and quantitatively analyze the impact of these uncertainties on ECG morphology and diagnostic markers. We analyze residual beat-to-beat variability in ECG recordings obtained from healthy subjects and patients. Using a biophysically detailed and anatomically accurate computational model of whole-heart electrophysiology combined with a detailed torso model calibrated to closely replicate measured ECG signals, we vary anatomical factors (heart location, orientation, size), heterogeneity in electrical conductivities in the heart and torso, and electrode placements across ECG leads to assess their qualitative impact on ECG morphology. Our study demonstrates that diagnostically relevant ECG features and overall morphology appear relatively robust against the investigated uncertainties. This resilience is consistent with the narrow distribution of ECG due to residual beat-to-beat variability observed in both healthy subjects and patients.

Figures (8)

• Figure 1: Left: Transmembrane potential $V_{\rm m}(\mathbf{x},t)$ for the reference simulation. Right: Lead field solutions $Z_i(\mathbf{x})$ considering the 9 electrodes position R, L, F, V1, V2, V3, V4, V5, and V6 to compute the 12-lead . The range of $Z_i(\mathbf{x})$ solutions, which is $[-1.0, 1.4]$, is restricted between $[0.5, 0.8]$ to improve the visualization.
• Figure 2: Top-left: cardiac model embedded in a spherical halo. For each investigated position of the heart, the halo is the only remeshed part of the torso. Bottom-left: schematic representation of the frontal, sagittal, and transverse plane of the torso. Top-right: schematic representations of the three directions of translation. Bottom-right: schematic representations of the three rotational axes.
• Figure 3: Distribution of the 12-lead (grey) and corresponding mean value (black) due to residual beat-to-beat variability of the signals. The recorded s refer to a representative healthy subject (top), AF patient (middle), and VT patient (bottom).
• Figure 4: Top: distribution of the 12-lead (grey) due to anatomical uncertainties introduced by varying position and orientation of the heart within the torso. The reference signal is also depicted (black). Bottom-left panel: maximum relative variation (in percentage) of the R, S, and T peak amplitudes caused by each transformation, for each lead. Bottom-right: absolute variation in R, S, and T peak amplitudes caused by each considered transformation, for each lead.
• Figure 5: Row 1-2: 12-lead obtained by scaling the heart of $\pm$10% of its original dimension with a prescribed activation sequence (row 1), and with prescribed conduction properties (row 2). Row 3: Maximum relative variation (in percentage) of the R, S, and T peak amplitudes for each lead, due to scaling of the heart when prescribing the activation sequence (left), and when prescribing the conduction properties (right).