An in-silico analysis of experimental designs to study right ventricular function and pulmonary hypertension

In-vivo studies of pulmonary hypertension (PH) have provided key insight into the progression of the disease and right ventricular (RV) dysfunction. Additional in-silico experiments using multiscale computational models have provided further details into biventricular mechanics and hemodynamic function in the presence of PH, yet few have assessed whether model parameters are identifiable prior to data collection. Moreover, none have used modeling to devise synergistic experimental designs. To address this knowledge gap, we conduct an identifiability analysis of a multiscale cardiovascular model across four simulated experimental designs. We determine a set of parameters using a combination of Morris screening and local sensitivity analysis, and test for identifiability using profile likelihood based confidence intervals. We employ Markov chain Monte Carlo (MCMC) techniques to quantify parameter and model forecast uncertainty in the presence of noise corrupted data. Our results show that model calibration to only RV pressure suffers from identifiability issues and suffers from large forecast uncertainty in output space. In contrast, parameter and model forecast uncertainty is substantially reduced once additional left ventricular (LV) pressure and volume data is included. A comparison between single point systolic and diastolic LV data and continuous, time-dependent LV pressure volume data reveals that even basic, functional data from the LV remedies identifiability issues and provides substantial insight into biventricular interactions. Author SummaryComputational models of cardiac dynamics are becoming increasingly useful in understanding the underlying mechanisms of disease. In-silico analyses are especially insightful in understanding PH and eventual RV dysfunction, as these conditions are diagnosed months to years after disease onset. Many researchers couple computational models with in-vivo experimental models of PH, yet few ever assess what data might be necessary or sufficient for parameter inference prior to designing their experiments. Here, we considered a multiscale computational model including sarcomere dynamics, biventricular interactions, and vascular hemodynamics, and assessed whether parameters could be inferred accurately given limited cardiac data. We utilized sensitivity analyses, profile likelihood confidence intervals, and MCMC to quantify parameter influence and uncertainty. We observed that RV pressure alone is not sufficient to infer the influential parameters in the model, whereas combined pressure and volume data in both the RV and LV reduced uncertainty in model parameters and in model forecasts. We conclude that synergistic PH studies utilizing computational modeling include these data to reduce issues with parameter identifiability and minimize uncertainty.