Berkeley Fluids Seminar
University of California, Berkeley
Bring your lunch and enjoy learning about fluids!
February 24, 2015
Dr. Mahdi Esmaily-Moghadam (Center for Turbulence Research, Stanford University)
Developing multiscale computational tools to make predictive-based improvement in clinical medicine
Abstract: Drawbacks associated with traditional trial-and-error approaches in the development of new surgical methods for cardiovascular disease, together with the exponential growth rate of computational power in the last few decades warrants exploration of alternative approaches based on predictive simulations. This transition, however, requires development of a new set of algorithms that incorporate high performance computational techniques into this emerging area. The current surgical procedures performed on stage one single ventricle (SV) heart patients is an excellent example of an existing clinical problem that has not been improved in the last five decades despite poor outcomes. Hence, with a focus on this surgery, I will describe a set of tools that we developed to efficiently simulate the circulatory system, enabling us to propose a novel alternate design to current surgical methods. I will discuss several methodological advances, ranging from an implicit multidomain modular coupling method to a novel preconditioned iterative algorithm for the parallel solution of the system of linear equations arising in the finite element solution of the Navier-Stokes equations. These methods are introduced with accuracy, stability, modularity, scalability, and efficiency considerations in mind. Then, employing these tools in the study of SV palliation, I will present results from optimization of the conventional surgery. I will then present a novel surgical approach, which may revolutionize the SV palliation pathway, as the main outcome of this study.