Mechanical and Civil Engineering Seminar
Computational and data-driven models for modern medicine
Abstract: Nano-scale transistors fill warehouse-scale supercomputers, yet their performance still constrains development of the medical therapies our lives depend upon. The challenge is no mystery: our physiology hosts a cornucopia of multi-scale and multi-phase mechanics, each requiring specialized methods to overcome resolution limitations. Recent developments in machine learning and data assimilation are poised to change this paradigm through judicious use of available computational resources. New data-driven models for cell-scale cardiovascular flows and therapeutic bubble cavitation developed in this light are presented. High-fidelity spectral boundary integral and diffuse-interface solvers inform model design and lead to state-of-the-art predictions. MFC, our open-source software, showcases these capabilities. Analysis of the large-scale simulations guides biomicrofluidic device design and burst-wave lithotripsy administration.
Contact: Sonya Lincoln at (626) 395-3385 email@example.com