The Best of Both: Toward a hybrid discrete and continuum multiscale platelet aggregation and coagulation model

点击翻译按钮获取中文摘要

This project brings together computational scientists and mathematical modelers to solve a fundamental multifaceted multiscale problem in human physiology, namely understanding the platelet aggregation and coagulation (PAC) processes that comprise blood clotting. Over the past two decades, mathematical biology experts, including the current investigators, have worked to apply sound mathematical modeling principles and computational methods to attempt to dissect the complex interactions that occur within the platelet aggregation and coagulation process - fluid-structure interactions, mechanical-chemical interactions, and structure-structure interactions, to name a few. The complexity of this problem is due to its multiscale nature in both space and time, the complex disparate physical and chemical processes involved, as well as the challenge of attempting to model processes for which experimental validation is difficult. Over recent years, the experimental world has made significant advances in accumulating data that might both help us model the PAC cascade more faithfully, and allow us to predict its pathological deviations. The challenge is in connecting these two worlds and this project has as its goal employed modern computing concepts (numerical and algorithmic) to meet this challenge. While this project focuses on the PAC cascade, it will also have impact in a wide range of multiscale, multidiscipline applications such as chemical engineering and material science.The physiological time scale for PAC is on the order of minutes. To date, the only model able to simulate the entire PAC process over such time scales is a meso-scale (continuum) model developed by co-PI Fogelson and collaborators. This capability comes at the cost of coarse-graining the geometry and mechanics of the PAC process. The co-PI has also developed a fine-grained model of platelet aggregation that is at the forefront of platelet modeling. However, while conceptually faithful to the mechanics of the PAC process and the geometric intricacies of the developing aggregates, the fine-grained model does not yet contain treatment of the chemical processes of coagulation. Further, this model is computationally expensive and can, in a reasonable amount of time, only simulate a small fraction of the physical time that the meso-scale model can. Consequently, the goals of this project are two-fold: first, to extend the simulation capabilities of the fine-grained model both in terms of conceptual fidelity to the PAC cascade, and also in terms of computational efficiency by implementation on hybrid (CPU/GPU) architectures; and second, to cross-validate in the multiscale context current meso-scale and the extended fine-scale PAC models. Accomplishing these goals is critical to our longer-term research objective of developing hybrid multiscale models - enabled by current and future experimental data - that combine the best features of both these models.