Supervisors

Suitable for

Abstract

Prerequisites: Computational Medicine (recommended)

 

Abstract

High-fidelity cardiac digital twins enable detailed investigation of individual anatomy, electrophysiology, mechanics, and therapy response. However, calibrating these personalised models is computationally demanding, especially when building large cohorts or updating models longitudinally as new clinical data become available. Recent advances suggest that calibration cost can be substantially reduced by exploiting shared structure across individuals and by constructing surrogate models that emulate the behaviour of full electromechanical simulations.

This project will explore Gaussian-process surrogates, emulator-based optimisation, and transfer-learning strategies to accelerate the calibration of patient-specific digital twins. The idea is to treat each calibrated model as part of a larger cohort, allowing information to be transferred across patients with similar anatomical, electrophysiological, or mechanical properties. Students will integrate surrogate models with an existing ventricular electromechanical modelling pipeline and evaluate how much computational cost can be reduced while maintaining physiological accuracy.

Possible research themes include:
(i) building Gaussian-process emulators for key electromechanical biomarkers;
(ii) designing cohort-aware calibration strategies that reuse model information across individuals;
(iii) developing methods for efficient recalibration of digital twins over time;
(iv) benchmarking surrogate-enabled methods against full high-fidelity calibration workflows.

The project will contribute to scalable digital-twin frameworks capable of supporting population studies, therapy testing, and longitudinal follow-up.

• References

[1] In-silico human electro-mechanical ventricular modelling and simulation for drug-induced pro-arrhythmia and inotropic risk assessment. https://doi.org/10.1016/j.pbiomolbio.2020.06.007

[2] Harnessing 12-lead ECG and MRI data to personalise repolarisation profiles in cardiac digital twin models for enhanced virtual drug testing. https://doi.org/10.1016/j.media.2024.103361