Applications of AI in Biomedical Engineering
- A deep learning model for efficient end-to-end stratification of thrombotic risk in left atrial appendage. We propose UDGCNN (U: U-net, DGCNN: Dynamic Graph Convolutional Neural Network) that can utilize data from the point cloud of patient-specific LAA geometries as inputs and predict multiple hemodynamic indexes for systematic assessment of the thrombotic risk.
- Transfer Learning on Physics-Informed Neural Networks for Tracking the Hemodynamics in the Evolving False Lumen of Dissected Aorta. We introduce warm-start PINNs (WS-PINNs), a computational framework leveraging transfer learning, physics modeling, experimental data, and neural networks, to address the limitations of existing approaches for hemodynamic analysis within the dissected aortas.
- TGM-Nets: A Deep Learning Framework for Enhanced Forecasting of Tumor Growth by Integrating Imaging and Modeling. We propose TGM-Nets, a deep learning framework that combines bioimaging and tumor growth modeling (TGM) for enhanced prediction of tumor growth. This proposed framework, developed based on physics-informed neural networks (PINNs), is capable of integrating the TGM and sequential observations of tumor morphology for patient-specific prediction of tumor growth.
- Artificial intelligence velocimetry and microaneurysm-on-a-chip for three-dimensional analysis of blood flow in physiology and disease. We present artificial-intelligence velocimetry (AIV) to quantify velocity and stress fields of blood flow by integrating the imaging data with underlying physics using physics-informed neural networks.
Applications of AI in Epidemiology Forecasting
- Physics-informed deep learning for infectious disease forecasting. We employ PINNs to leverage the compartment models inspired by the underlying dynamics of the infectious disease and the observable data to enhance the capability of infectious disease forecasting.
Multiscale Mechanistic Modeling of RBC Disorders and Thrombosis
We have developed multiscale models based on physics laws using various numerical methods, such as molecular dynamics (MD), dissipative particle dynamics (DPD) and spectral element method (SEM), to simulate biological systems that span multiple spatial scales, including molecular level, protein level, sub-cellular level, cellular level, multi-cell systems, vasculature and organ systems. These studies have demonstrated that multiscale modeling can bridge the gap between the microscopic and macroscopic physiological processes, and provide innovative approaches to study key problems in biology, medicine and biomedical engineering, such as building mechanistic models to investigate the pathogenesis of human diseases and developing predictive models to test the existing hypotheses and derive new hypotheses to steer experimental and computational studies. These computational models also can be used to examine the efficacy of medication and facilitate the development of precision medicine, illustrate critical biological processes that cannot be observed through in vivo or in vitro experimental methods, thereby providing insights for development of new disease treatment strategies. Selected work is summarized in the figure below.
