Identifying Dominant Parameters in PV-based Fitting of a Paediatric Cardiac Digital Twin

Ana Montañés1, Jorge Sanchez2, Ferran Prats-Domenech3, Ferran Roses-Noguer3, Javier Saiz4, Beatriz Trenor5
1Centro de Investigación e Innovación en Bioingeniería (Ci2B), Universitat Politècnica de València, 2Center for Research and Innovation in Bioengineering (Ci2B), Universitat Politecnica de Valencia, 3Hospital Vall d'Hebron, 4Centro de Investigacion e Innovacion en Bioingenieri­a (Ci2B), Universitat Politecnica de Valencia, 5Universitat Politècnica de València


Abstract

Aims: Closed-loop cardiac electromechanical models are increasingly used to support in silico experimentation and digital twins by coupling myocardial mechanics with circulatory dynamics. However, their personalization remains challenging under limited clinical data, particularly in paediatric applications. In this setting, the inverse problem becomes underdetermined, as multiple parameter combinations can reproduce similar pressure-volume (PV). In this work, we aim to identify which parameters influence the model to reproduce physiologically PV loops for developing a closed-loop electromechanical paediatric digital twins.

Methods: We consider a whole-heart patient-specific closed-loop cardiac electromechanical model derived from the MRI scan of a 9-year-old subject. Personalization is performed through myocardial material parameters, governed by a Usyk-type constitutive formulation, including ventricular parameters controlling the stiffness of the passive properties of the myocardium (aScaleLV, bScaleLV, aScaleRV, bScaleRV), and circulatory parameters including systemic arterial resistance and compliance (SysArtResist, SysArtCompli) and venous compliance (SysVenCompli). A parameter screening study is conducted on a set of available simulations, and parameter influence is quantified using Morris elementary effects, with the mean absolute effect used for ranking.

Results: The PV screening revealed a clear dominance of left ventricular parameters. aScaleLV showed the largest influence across all observations, with a Morris index value of 13415.27, followed by SysArtResist (2066.5) and SysArtCompli (273.83).

Conclusion: PV calibration overemphasizes left ventricular parameters, with reduced sensitivity to vascular and right ventricular parameters.