Investigating the Progression of Aortic Disease with Mechanics-driven Agent-Based Modelling

Agent-based modelling (ABM) framework is increasingly being used in investigating the cause and modelling the progression of cardiovascular diseases including atherosclerosis, in-stent restenosis, and vein graft adaptation. Unlike equation-based modelling, ABM can capture interactions between individual elements within complex systems by prescription of biological rules (sometimes probabilistic) to each agent, which can be cells or an aggregation of similar cells. This non-parametric and statistics-driven approach means ABM is more equipped than continuous models to tackle heterogeneous, discrete, and stochastic problems, such as aorta morphological evolution. However, one major challenge in scaling ABMs, especially in a cardiovascular context, is coupling this method with existing interpretable modelling methods such as finite element analysis (FEA) or computational fluid dynamics (CFD). More specifically, it is difficult to incorporate the changes in aorta morphology predicted by the ABM framework at each timestep to FEA and CFD where turbulence analysis and stress field prediction can be performed. It is also equally difficult to incorporate the changes of stress and turbulence field at each timestep due to changes in aorta morphology into the ruleset for the ABM framework.

Our work addresses this problem by building a model that can accept time-varying attributes related to FEA and CFD work such as stress and turbulence. Using FEA and CFD results on test cases, we build a relation between aorta morphology and stress and turbulence that can be more easily applied to the ABM framework. This approach reflects changes caused by the two-way interaction between cellular response and tissue variables such as stress and turbulence over simulation time while reducing computational costs of running FEA and CFD at each timestep. Finally, the model is validated and calibrated with established datasets and is used to predict the evolution of aorta morphology in clinically sourced data points.