Statistical, Mathematical and Computational Modelling for Personalised Cardiovascular Healthcare

Harry Saxton and Xu Xu (University of Sheffield)

Wednesday 24th April 14:00-15:00 Maths 311B

Abstract

Part 1 (given by Harry Saxton)

Sensitivity analysis and parameter orthogonality work for 0D cardiovascular circulation models

Part 2 (given by Xu Xu)

Multi-component lattice Boltzmann method to simulate 3D haemodynamics, with potential clinical applications.

 

More details:

 

Mathematical and computational models are increasingly viewed as important tools to support clinical diagnosis and decision making for cardiovascular healthcare.

 

In the first half of this seminar, Harry will present our work on input parameter identification from lumped-parameter blood circulation models, or, within the clinical context, the cardiovascular personalisation process. We use various statistical techniques to analyse the complex high-dimensional input parameter hyperspace associated with the class of models. We begin by examining sensitivity indices, assessing their convergence to pinpoint the most influential parameters. Additionally, we introduce a novel global subset selection method to identify clinically significant biomarkers and to enhance model personalisation accuracy. In parallel, we utilise Sobol indices to propose a domain-agnostic, intuitive approach for pseudo-mapping of the input hyperspace. These findings present a global methodology for input parameter identifiability and hyperspace mapping, offering valuable insights into streamlining cardiovascular model personalisation.

 

In the second half, Xu will present a recently developed (from James Spendlove’s PhD, completed in 2022) three-dimensional multicomponent lattice Boltzmann method, capable of simulating vesicles (such as erythrocytes) at cellular scale. The presented method is encapsulated in a single framework, where the application of the immersed boundary force in the automatically adaptive interfacial region results in correct vesicle behaviour. The benefits of such a model are its transparent methodology, stability at high levels of deformation, automatic-adaptive interface, and potential for the simulation of many erythrocytes. We demonstrate the utility of the model by examining the steady-state properties, as well as dynamical behaviour within shear flow. The model stability is highlighted through handling of large deformations and vesicle interactions, opening future haemodynamic applications in simulating cell interactions in stenosed or stented arteries and deformed erythrocytes passing through capillaries. 

 

 

 

Add to your calendar

Download event information as iCalendar file (only this event)