NSF Award
2436630
A digital twin is a virtual model that mirrors and updates in real-time based on data from its physical counterpart. In biomedical and healthcare fields, digital twins, representing virtual models of patients, medical devices, and more, can open up new avenues for developing and evaluating innovative biomedical technologies, particularly enabling virtual clinical trials for evaluating cardiovascular medical devices and advancing regulatory sciences. However, current digital twin technologies lack sufficient computational fidelity and efficiency to effectively support these biomedical and healthcare applications. To resolve these challenges, this project aims to develop advanced computational methods for creating high-fidelity, fast-running digital twins of patient hearts and cardiovascular medical devices. Additionally, the methods will be made publicly available through a software/cyberinfrastructure platform. This will facilitate virtual clinical trials that can evaluate the efficacy and safety of medical devices, as well as improve device designs before initiating real clinical trials in a safe, cost-effective, and precisely controlled manner. In addition to advancing digital twin technologies, the project’s cyberinfrastructure will serve as an educational resource for students, researchers, and industrial engineers to enhance their understanding of advanced digital twin techniques for medical device evaluation.<br/><br/>This project will develop novel machine learning (ML)-based image analysis algorithms and physics solvers for performing near-realtime virtual clinical trials with high-fidelity digital twins of patient hearts and cardiovascular medical devices. Patient-specific geometries and tissue mechanical properties will be incorporated into the digital twin construction for near-realtime physics simulations. Consequently, virtual clinical trials can be performed at significantly reduced time and financial costs. This project will deliver (1) novel ML algorithms for accurate digital twin geometry reconstruction from 3D+t medical images, enabling point-to-point mesh correspondence for high-fidelity dynamic motion tracking; (2) a robust and computationally efficient inverse method to identify in vivo material properties from medical images, which is essential for creating material-realistic digital twins; (3) a new ML-based fluid-structure interaction (ML-FSI) solver for biomechanics and hemodynamic analyses, thereby enabling dynamic digital twin simulations throughout a cardiac cycle. While the primary focus will be on digital twins of the left heart and aorta, the computational methods can be generally applied to create digital twins of the entire heart. The computational methods will be demonstrated through concrete examples involving Transcatheter Aortic Valve Replacement (TAVR) and Thoracic Endovascular Aortic Repair (TEVAR) devices. The algorithms and methods developed in this project will be generic and readily applicable to devices for treating various cardiovascular diseases.<br/><br/>This project is jointly funded by the Division of Mathematical Sciences, the OAC Cyberinfrastructure for Sustained Scientific Innovation (CSSI) program, and the CBET Engineering of Biomedical Systems program.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
