Patient‐Specific Lumped‐Parameter Model for Quantifying Vessel‐Specific Remodeling and Predicting Right Ventricular Function in Pulmonary Hypertension

TL;DR

This study creates a personalized model to assess right ventricular function and distinguish PH subtypes using routine clinical data.

Contribution

A patient-specific lumped-parameter model is developed to quantify vessel-specific remodeling and simulate RV function in PH.

Findings

The model showed good agreement with clinical data across PH phenotypes.

Model-derived parameters highlighted physiological distinctions among PH subtypes.

Predicted RV PV loops revealed phenotype-specific differences in volumes and pressures.

Abstract

Pulmonary hypertension (PH) is a heterogeneous disease with patient‐specific variability and vessel‐specific remodeling, which eventually lead to right ventricular (RV) failure. The gold standard for RV assessment—pressure–volume (PV) loop acquisition—is invasive and limited to specialized settings. This study aims to develop a patient‐specific lumped‐parameter model that quantifies vessel‐specific remodeling and simulates RV PV loops across PH phenotypes using routine clinical data.

A lumped‐parameter model was calibrated using right heart catheterization and echocardiography data. Model agreement was assessed by R
2 values for pressure and flow goodness‐of‐fit, and model‐derived hemodynamic metrics were compared with clinical values. A dimensionality reduction approach was applied to investigate how well different PH phenotypes could be separated.

Across the cohort, the lumped‐parameter model showed good agreement with clinical data. Model‐derived vessel‐specific (pulmonary arterial, capillary, venular) parameters highlighted physiological distinctions among phenotypes. Predicted RV PV loops revealed phenotype‐specific differences in right ventricular volumes, pressures, and stroke work. The linear discriminant analysis (LDA) demonstrated qualitative separability, indicating that model‐derived, nonmeasurable features offer additional discriminatory information.

Our results demonstrate that lumped‐parameter models can be calibrated to clinical data to quantify vessel‐specific remodeling and simulate RV pressure–volume dynamics to provide useful information for distinguishing among different PH phenotypes. This underscores the potential of computational models as noninvasive, clinically feasible tools for assessing in‐depth pulmonary vascular and RV function in PH.

Personalized, patient‐specific biophysical model enabling mechanistic assessment of right ventricular (RV) function as well as robust differentiation of pulmonary hypertension (PH) subtype from routine clinical measurements of pulmonary arterial pressure and flow.

Figures