Forward uncertainty quantification and sensitivity analysis in models of systemic circulation

Abstract

The intricate nature of the heart and blood circulation is intensively studied in the search for answers and insights capable of maturing the understanding of the cardiovascular system’s physiological and pathophysiological phenomena. Cardiovascular computational models are useful tools for this purpose. They are already widely used by the medical-scientific community, simulating important phenomena such as the dynamics of the systemic circulation and providing valuable information, such as hemodynamic parameters and biomarkers, of common clinical use. However, the clinical application of these models is not straightforward, and for them to be used ubiquitously for decision-making, there is still much to be improved. An important step in this direction is to search for more accurate and reliable models, where the understanding of the relationship between the uncertainties in the input parameters of a model and the precision of its results must be taken into account. In the present work, we verify the effect of the propagation of uncertainties on the input parameters of lumped parameter models and a multi-scale finite element model that simulates the systemic circulation dynamics. For this, we perform forward uncertainty quantification and sensitivity analysis based on the polynomial chaos expansion. The results obtained point to the most influential parameters in the prediction of quantities of interest of clinical relevance. Thus, it is expected that the knowledge acquired on the parameters that must be measured with greater precision and the least influential ones, which can be measured from population-based values or the literature, can help in the calibration and development of more accurate and consistent models.