Our computational lab at Duke University develops advanced computational frameworks to transform how we understand, monitor, and treat human disease. We design large-scale, high-performance applications that bridge numerical modeling, biomedical simulation, and clinical translation.
A central focus is building multiscale, 3D vascular digital twins from CT or MRI imaging. These models integrate our Adaptive Physics Refinement (APR) framework—capturing subcellular interactions such as rare-cell transport—with HARVEY, our massively parallel blood-flow solver capable of simulating millions of red blood cells at organ and whole-body scales. Using the Longitudinal Hemodynamic Mapping (LHM) framework, we extend these simulations over millions of heartbeats, enabling time-resolved predictions of disease progression and treatment outcomes.
By combining physics-based modeling with AI and machine learning, we accelerate simulation, discover new digital biomarkers, and support real-time, personalized decision making. Applications span cardiovascular disease diagnosis, carotid plaque monitoring, heart-failure management, and cancer cell migration in the bloodstream.
The image below provides an overview of multiscale vascular digital twin modeling: Left – APR tracking subcellular interactions; Middle – HARVEY simulation of millions of red blood cells; Right – aortic arch velocity streamlines from a patient-specific HARVEY model.
