[Paper Review] Cardiac Digital Twin Pipeline for Virtual Therapy Evaluation

Julià Camps, Zhinuo Jenny Wang+7 more|arXiv (Cornell University)|Jan 18, 2024

ECG Monitoring and AnalysisMedicine3 citations

TL;DR

This paper presents an automated pipeline for creating patient-specific cardiac digital twins using routine CMR and 12-lead ECG data, employing a reaction-Eikonal model with Purkinje network and ionic dynamics to simulate electrophysiology. It achieves high ECG fidelity (r = 0.93) and enables virtual therapy evaluation with dose-dependent QT prolongation matching clinical data.

ABSTRACT

Cardiac digital twins are computational tools capturing key functional and anatomical characteristics of patient hearts for investigating disease phenotypes and predicting responses to therapy. When paired with large-scale computational resources and large clinical datasets, digital twin technology can enable virtual clinical trials on virtual cohorts to fast-track therapy development. Here, we present an automated pipeline for personalising ventricular anatomy and electrophysiological function based on routinely acquired cardiac magnetic resonance (CMR) imaging data and the standard 12-lead electrocardiogram (ECG). Using CMR-based anatomical models, a sequential Monte-Carlo approximate Bayesian computational inference method is extended to infer electrical activation and repolarisation characteristics from the ECG. Fast simulations are conducted with a reaction-Eikonal model, including the Purkinje network and biophysically-detailed subcellular ionic current dynamics for repolarisation. For each patient, parameter uncertainty is represented by inferring a population of ventricular models rather than a single one, which means that parameter uncertainty can be propagated to therapy evaluation. Furthermore, we have developed techniques for translating from reaction-Eikonal to monodomain simulations, which allows more realistic simulations of cardiac electrophysiology. The pipeline is demonstrated in a healthy female subject, where our inferred reaction-Eikonal models reproduced the patient's ECG with a Pearson's correlation coefficient of 0.93, and the translated monodomain simulations have a correlation coefficient of 0.89. We then apply the effect of Dofetilide to the monodomain population of models for this subject and show dose-dependent QT and T-peak to T-end prolongations that are in keeping with large population drug response data.

Motivation & Objective

• To develop an automated pipeline for generating personalized cardiac digital twins from routinely acquired clinical imaging and ECG data.
• To enable virtual clinical trials by simulating patient-specific electrophysiological responses to therapies.
• To quantify parameter uncertainty in digital twins through population-based inference rather than single-model estimation.
• To translate fast reaction-Eikonal simulations into more realistic monodomain simulations for improved physiological fidelity.
• To validate the pipeline’s predictive capability using drug response data, particularly for Dofetilide-induced QT prolongation.

Proposed method

• Construct patient-specific ventricular anatomy from cardiac MRI (CMR) data using segmentation and mesh generation.
• Apply sequential Monte-Carlo approximate Bayesian computation (SMC-ABC) to infer electrical activation and repolarisation parameters from 12-lead ECG.
• Use a reaction-Eikonal model with biophysically detailed ionic currents and Purkinje network to simulate action potential propagation.
• Generate a population of models to represent parameter uncertainty, enabling uncertainty propagation in therapy evaluation.
• Develop a translation framework from reaction-Eikonal to monodomain simulations for enhanced realism in electrophysiological dynamics.
• Calibrate and validate simulations against clinical ECG recordings and known drug response patterns.

Experimental results

Research questions

• RQ1 Can a fully automated pipeline generate patient-specific cardiac digital twins using only standard CMR and 12-lead ECG data?
• RQ2 To what extent can reaction-Eikonal simulations reproduce individual patient ECGs with high fidelity?
• RQ3 How accurately can the pipeline predict dose-dependent drug effects on repolarization, such as QT and T-peak to T-end prolongation?
• RQ4 Can uncertainty in model parameters be effectively captured and propagated through virtual therapy evaluation?
• RQ5 How well do translated monodomain simulations compare to reaction-Eikonal results in reproducing clinical ECG features?

Key findings

• The pipeline achieved a Pearson correlation coefficient of 0.93 between simulated and patient-ECG in a healthy female subject using the reaction-Eikonal model.
• Monodomain simulations, derived from the reaction-Eikonal population, reproduced the ECG with a correlation coefficient of 0.89.
• Dofetilide administration in the monodomain model population induced dose-dependent QT prolongation consistent with large-scale clinical drug response data.
• T-peak to T-end interval prolongation was also observed in a dose-dependent manner, aligning with known pharmacological effects.
• The use of a population of models instead of a single model effectively captured parameter uncertainty and enabled robust therapy evaluation.
• The translation from reaction-Eikonal to monodomain simulations preserved key electrophysiological features while enhancing biological realism.