Developing a Digital Twin of the Cardiopulmonary System in a Mouse: Inferring Hemodynamics from Sparse Measurements

Vitaly O. Kheyfets, Kenzo Ichimura, Paul M. Heerdt, Mengqian Zhang, Ella Lyon, Kurt R. Stenmark, Edda Spiekerkoetter
University of Colorado School of Medicine and University of Colorado Anschutz Medical Campus. Stanford University. Yale School of Medicine.
United States

Annals of Biomedical Engineering
Ann Biomed Eng 2026;
DOI: 10.1007/s10439-026-04107-8

Abstract
Purpose: The use of rodent models in the study of cardiopulmonary disease is widespread, but comprehensive functional and hemodynamic characterization of the cardiopulmonary system in each rodent is often impractical and may require integration of multimodal measurements. The objective of this study is to evaluate a 0D cardiopulmonary model for simulating mouse-specific physiology and inferring individualized parameters.
Methods: We developed a 0D model of the cardiopulmonary axis, bounded by the right and left atria, incorporating 13 unknown parameters representing vascular impedance, RV pressure (RVP)-volume dynamics, and tricuspid/pulmonic valve regurgitation. The model fitted RVP and volume data from 28 mice across four surgical conditions, including two scenarios of mechanically induced RVP overload. Sensitivity and identifiability analyses revealed a reduced subset of nine parameters that were structurally and practically identifiable.
Results: Optimization of the identifiable parameters adequately reproduced RVP waveforms (r = 0.94 for maximum dP/dt with LOA < 1 mmHg s-1) and volume extrema (r = 0.97 for EDV, r = 0.98 for ESV with LOA for both measurements ~ 5 μL), while one non-physiological case was excluded from the analysis. As expected, mice with RVP overload exhibited elevated inferred Ees, Eed, and RAP. Also, model-inferred Ees was moderately correlated with single-beat estimates of RV contractility (r = 0.68, p < 0.01).
Discussion: This study demonstrates that subject-specific computational modeling enables inference of ventricular function and pulmonary hemodynamics from RV pressure and volume data. This approach provides access to otherwise unmeasurable quantities and lays the groundwork for digital twins to support disease tracking and in silico testing of interventions.

Category
Animal Models of Pulmonary Vascular Disease and Therapy
Mechanical and Computer Models of Pulmonary Vascular Disease and Therapy

Age Focus: No Age-Related Focus

Fresh or Filed Publication: Fresh (PHresh). Less than 1-2 years since publication

Article Access
Free PDF File or Full Text Article Available Through PubMed or DOI: No

Scroll to Top