Physics-Constrained Modeling and Optimization of Complex Systems: Healthcare Application

Jianxin Xie

Rapid advances in sensing and imaging techniques have created a data-rich environment and tremendously benefited data-driven predictive modeling and decision-making for complex systems. Realizing the full potential of the sensing and imaging data depends on the development of novel and reliable analytical models and tools for system informatics. The goal of my research is to develop innovative physics-augmented methodologies for modeling, monitoring, and optimizing complex systems. In this talk, I will present two topics to tackle the challenges in complex systems modeling and optimization. In the first topic, a physics-constrained deep learning method is developed to model the spatiotemporal inverse systems. This method integrates physics-based principles with the advanced deep learning infrastructure to predict the spatiotemporal system dynamics based on indirect and noisy sensor observations. This methodology is implemented in inverse electrocardiography (ECG) modeling, which generates a robust prediction of electrical potential mappings on the heart surface based on body-surface sensor measurements. In the second topic, a novel physics-augmented strategy is proposed for optimal sensor placement to actively explore and model the dynamics of 3D complex-structured systems. This active learning scheme not only combines uncertainty estimation and space-filling design over the complex geometry but also respects the underlying physics-based prior knowledge, enabling effective learning of system dynamics from limited sensor exploration. The framework was implemented to estimate the electrodynamics in both healthy and diseased 3D cardiac systems. The proposed frameworks have profound potential in modeling other spatiotemporal multi-sensor systems.

Jianxin Xie received her BS in Physics from Southeast University, China, and her MS from Florida State University, USA. She is currently pursuing a PhD with the Department of Industrial Systems Engineering, University of Tennessee Knoxville. Her current research interests lie in advanced data analytics, data mining, and physical-statistical modeling with applications in various complex systems. She is the recipient of several awards such as IISE Gilbreth Memorial Fellowship, QCRE and DAIS Track Best Poster Award, and Runner-up Best Paper Award in Operation Research Track in the 2021 IISE Annual Conference. Jianxin is a student member of IISE, INFORMS, and IISE.