Multi-Fidelity Modeling and Uncertainty Quantification of the Coagulation Cascade in Patient-Specific Left Atrial Flows
Please login to view abstract download link
Thrombosis, a complex physiological process involving platelet aggregation and the coagulation cascade, is often modeled using numerous 3D advection-reaction-diffusion partial differential equations (PDEs) in the low-diffusivity regime. This computational challenge is addressed through the development of a Multi-Fidelity (MuFi) approach, transforming 3D PDEs into ordinary differential equations (ODEs) and substantially reducing computational costs. The study applies the MuFi approach to patient-specific left atrium (LA) models, assessing uncertainty in species' initial concentrations. Utilizing a database of LA models, including thrombus-negative and positive cases, a brute-force Monte Carlo method is employed for uncertainty quantification. The focus is on thrombin concentration in the left atrial appendage (LAA), a critical site for clot formation, showcasing the computational efficiency and applicability of the MuFi model in understanding thrombosis in patient-specific contexts.
