Effects of arterial wall models and measurement uncertainties on cardiovascular model predictions

• Published in: Journal of biomechanics 2017-01, Vol.50, p.188-194
• Main Authors: Eck, V.G, Sturdy, J, Hellevik, L.R
• Format: Article
• Language: English
• Subjects: Algorithms; Aorta; Arterial Pressure; Arterial wall models; Arteries; Blood flow; Blood pressure; Brachial Artery - physiology; Calibration; Cardiovascular diseases; Classification; Compliance; Data processing; Experimental data; Femur; Flow simulation; Fluid dynamics; Health risks; Heart; Hemodynamics; Humans; Hydrodynamics; Integration; Limbs; Mathematical models; Mechanical properties; Methods; Models, Cardiovascular; Parameter estimation; Parameter sensitivity; Parameter uncertainty; Patient-Specific Modeling; Physical Medicine and Rehabilitation; Predictions; Propagation; Scaling; Sensitivity analysis; Simulation; Travel; Uncertainty; Uncertainty quantification; Veins & arteries; Velocity; Wave propagation; Wave propagation model; Wave velocity
• ISSN: 0021-9290; 1873-2380
• DOI: 10.1016/j.jbiomech.2016.11.042

Abstract We developed a methodology to assess and compare the prediction quality of cardiovascular models for patient-specific simulations calibrated with uncertainty-hampered measurements. The methodology was applied in a one-dimensional blood flow model to estimate the impact of measurement uncertainty in wall model parameters on the predictions of pressure and flow in an arterial network. We assessed the prediction quality of three wall models that have been widely used in one-dimensional blood flow simulations. A 37-artery network, previously used in one experimental and several simulation studies, was adapted to patient-specific conditions with a set of three clinically measurable inputs: carotid–femoral wave speed, mean arterial pressure and area in the brachial artery. We quantified the uncertainty of the predicted pressure and flow waves in eight locations in the network and assessed the sensitivity of the model prediction with respect to the measurements of wave speed, pressure and cross-sectional area. Furthermore, we developed novel time-averaged sensitivity indices to assess the contribution of model parameters to the uncertainty of time-varying quantities (e.g., pressure and flow). The results from our patient-specific network model demonstrated that our novel indices allowed for a more accurate sensitivity analysis of time-varying quantities compared to conventional Sobol sensitivity indices.