Randles Lab

I came to Duke as an undergraduate, spent more than a few nights in K-ville, graduated, left for a few years, and then had the opportunity to return as faculty in Biomedical Engineering. This month I was promoted to Full Professor. And yes, it's pretty fun that my title is now simply "Professor." I'm incredibly grateful to the mentors who believed in me, the students, postdocs, staff, and collaborators who make our research possible, to my colleagues for their support and friendship over the years, and to my family for supporting me through every stage of this journey. I'm excited for what comes next and grateful that I get to continue building it here at Duke.

📚 CompHealth Corner: What We're Reading Understanding how circulating tumor cells interact with blood vessel walls is critical to understanding cancer metastasis—but capturing these interactions across realistic vascular networks at submicrometer resolution has historically required prohibitive computational resources. This week, we're highlighting a new publication from Center Director Amanda Randles and Center members Aristotle Martin, William Ladd, and Runxin Wu in the Journal of Computational Science. Their work introduces High-Throughput Adaptive Physics Refinement (APR-AD), a computational framework that dramatically expands the scale of adhesive transport simulations while maintaining the detailed physics needed to model receptor-mediated interactions. By combining adaptive physics refinement with hybrid CPU-GPU computing and scalable parallel algorithms, the platform can simulate more than 3,000 circulating tumor cells simultaneously while reducing memory requirements by approximately 15× compared to fully explicit models. These advances transform APR-AD into a powerful computational microscope for studying cancer transport and other biological processes that span multiple length scales. At the Center for Computational and Digital Health Innovation, we're proud to see our researchers continue pushing the boundaries of computational science, enabling simulations that bring us closer to understanding—and ultimately combating—complex diseases like cancer. Read more: comphealth.duke.edu/publications/h…

I'm excited to share our latest publication in IEEE Computing in Science & Engineering: "Retrospective on the Lax Report: Then and Now" It was an honor to work alongside an outstanding group of leaders in high performance computing: Ewa Deelman, Jack Dongarra, Bruce Hendrickson, Dan Reed, Edward Seidel, and Kathy Yelick to reflect on the remarkable legacy of the 1982 Lax Report and discuss what is needed to sustain scientific computing over the next four decades. The article examines how HPC has evolved from centralized supercomputers to today's AI-enabled, cloud-augmented ecosystem while highlighting the continued importance of scientific software, hardware-software co-design, workforce development, and sustained public investment. As AI reshapes computing, ensuring that future architectures continue to meet the needs of rigorous scientific simulation is more important than ever. 📄 Read the article: computer.org/csdl/magazine/…

I'm honored to have authored the accompanying editorial in the @NEJM discussing the scientific foundations behind two important clinical trials evaluating computationally enabled, angiography-derived physiological assessment for coronary artery disease. For many years, my research group has worked toward the vision that high-fidelity computational modeling can help bring physiology directly into clinical decision-making. These studies represent another important step toward that vision, demonstrating how computational methods can make physiology-guided care more accessible while highlighting the continued convergence of computational science, biomedical engineering, and cardiovascular medicine. It has been a privilege to reflect on the evolution of this field and to discuss where it may be headed next. Congratulations to the investigators on these important studies, and thank you to the New England Journal of Medicine for the opportunity to contribute the accompanying editorial. #Cardiology #ComputationalScience #DigitalHealth #DigitalTwins #CoronaryArteryDisease #FFR #BiomedicalEngineering

It's a busy conference week for the Randles Lab! After presenting one paper at ICCS 2026 in Hamburg, Ayman Yousef is heading straight to PASC 2026 to present another paper on Wednesday, July 1. Title: "GPU Halo Replay: Lossless Twin Simulations for Flexible In Situ Analysis of Stencil-Based Solvers." This work, by Ayman Yousef, Aristotle Martin, Ph.D., and I, introduces GPU Halo Replay, a framework that enables lossless halo replay so analysis can be performed by dedicated computational twin simulations without slowing the primary simulation or sacrificing fidelity. If you're attending PASC and are interested in HPC, in situ analysis, scientific workflows, or scientific visualization, he would love to see you at the talk and discuss the paper! #PASC2026 #ICCS2026 #HPC #InSitu #ScientificComputing #Visualization #CFD

@SanSanychUA great question! I think it's really starting to rival compute. Data quality, provenance, and trust are becoming top priorities. We can build incredibly fast simulations, but if we can't trust the data flowing through the workflow, we can't trust the digital twin.

@RandlesLab "Digital twins are workflow problems" is exactly right. In industrial settings the bottleneck is often HPC-grade simulation meeting edge telemetry latency, not the 3D model itself. Curious: do you see orchestration/data lineage as the bigger blocker than compute?

The Randles Lab is heading to Hamburg for ICCS 2026! This year we'll be presenting research across longitudinal digital twins for cardiovascular disease, AI-enabled HPC simulation workflows, and multiscale modeling of cancer cell transport. From reconstructing physiology over millions of heartbeats to understanding how red blood cell heterogeneity influences circulating tumor cell arrest, our work is focused on building predictive computational models for human health. Looking forward to great discussions throughout ICCS. If you'll be there, come say hello! Conference link: iccs-meeting.org/iccs2026/ #ICCS2026 #HPC #DigitalTwins #AI #ComputationalScience #CancerResearch #Hemodynamics #Exascale

One of my favorite sessions at ISC High Performance this week was on The Role of AI and the HPC-Edge Continuum in Digital Twin Workflows with Peter Coveney and Anuj Kapadia. A takeaway that kept coming up is that digital twins are not just simulation problems. They're entire data workflow and ecosystem problems. Getting a single simulation to run faster is important, but that's only one piece. To make digital twins practical at scale, we need the entire ecosystem: VVUQ, data management, orchestration and scheduling, provenance and data lineage, monitoring and failure recovery, reproducibility, automation, and, of course, the underlying physics. "Digital twins are workflow problems" #ISC2026 #digitaltwins

Looking forward to speaking today at the 2nd Advancing Autonomous Scientific Discovery (A2SD) Workshop at ISC High Performance 2026 in Hamburg. My talk, "Chronophysiomics: Autonomous Discovery Through Longitudinal Digital Twins," explores how longitudinal digital twins can combine wearable data, medical imaging, electronic health records, and physics-based simulation to model cardiovascular physiology across millions of heartbeats. As AI and high-performance computing continue to advance, the opportunity is no longer just building digital twins—it is enabling trustworthy, autonomous scientific systems that continuously learn from data, reason through physics, and accelerate discovery. I'm excited to discuss how these ideas can help move healthcare from reactive to predictive and ultimately proactive care, and to learn from the outstanding speakers and attendees throughout the workshop. Workshop details: lnkd.in/epx7u85p #ISC2026 #A2SD2026 #DigitalTwins #AutonomousScience #AI #HighPerformanceComputing #BiomedicalEngineering #DigitalHealth #CardiovascularResearch #Chronophysiomics

Can we forecast human health the way we forecast weather? I'll be exploring that question in my ISC High Performance 2026 midweek keynote, HPC for Vascular Digital Twins, in Hamburg. The talk will explore how advances in high performance computing, AI, and physics-based modeling are enabling personalized digital twins that can track physiology across millions of heartbeats, forecast treatment outcomes, and move healthcare from reactive to proactive. If you're attending ISC, I'd love to see you there!

Surprise: Chinese LineShine supercomputer takes the #1 spot on the #Top500 and #HPCG lists! First time since a while that spot is taken by a homogeneous CPU machine. So is heterogeneity here to stay? CPUs with HBM and tensor cores (e.g., ARM SME). Congrats Yutong and folks.

I enjoyed the opportunity to discuss how advances in high-performance computing and AI are helping make healthcare digital twins a reality. From modeling blood flow in patient-specific vascular networks to understanding physiology over millions of heartbeats, we're working toward a future where computational models can help identify risk earlier, test interventions virtually, and support more proactive healthcare. Thank you to Scientific Computing World for the feature. scientific-computing.com/article/scw75-…

We are pleased to share that Amanda Randles and Wentao Ma from the Randles Lab, together with Xuning Zhao and Mauro Rodriguez Jr. from Brown University, have co-organized the mini-symposium “Computational Hemodynamics and Cardiovascular Mechanics ” at USNC-TAM26, the 20th U.S. National Congress on Theoretical and Applied Mechanics. The mini-symposium will be held on June 22 from 9:20 a.m. to 12:40 p.m. at the Pasadena Convention Center. Held every four years, this prestigious conference brings together leading researchers, educators, and practitioners from across the fields of theoretical and applied mechanics to exchange ideas, present emerging discoveries, and discuss future directions in mechanics research, education, and real-world applications. As part of the symposium, Dr. Ma from the Randles Lab will also be presenting a talk titled “An efficient and scalable immersed boundary framework for cardiovascular flow in compliant vessels” The symposium will highlight recent advances and ongoing challenges in areas including biomechanics, computational modeling, and cardiovascular simulation. We encourage researchers, students, and professionals interested in these fields to attend USNC-TAM26 (usnctam26.org) and participate in this exciting mini-symposium. We look forward to the engaging discussions and scientific exchange it will foster. #BME #DigitalTwins #BiomedicalEngineering #CardiovascularHealth Duke University Pratt School of Engineering Duke University

Congratulations to Randles Lab Ph.D. student Nusrat Sadia Khan on receiving Duke University's Outstanding Teaching Assistant Award! 🎉 Recognized for her exceptional teaching, mentorship, and dedication to student success in BME 307: Transport Phenomena in Biological Systems. #DukeBME

📚 CompHealth Corner: What We're Reading As artificial intelligence becomes increasingly integrated into health care, questions surrounding privacy, transparency, and data ownership are more important than ever. This week, we're highlighting a recent Viewpoint in The Lancet Digital Health co-authored by our Associate Director of Wearables, Jessilyn Dunn, which examines the growing commercialization of health data and the need for stronger safeguards around how patient information is shared, used, and monetized. The authors call for clearer health data transaction disclosure requirements, patient-centered data practices, and regulatory frameworks that promote both innovation and accountability. As digital health technologies continue to evolve, building and maintaining public trust will be essential to ensuring that data-driven advances benefit patients while protecting privacy and individual rights. At the Center for Computational and Digital Health Innovation, we believe these conversations are critical to shaping the future of responsible digital health. 🔗 Read the full article: comphealth.duke.edu/publications/t…

🎉 We are proud to share that Nusrat Sadia Khan from Randles Lab has been honored with the Outstanding Teaching Assistant Award in recognition of her exceptional commitment and contribution to teaching the BME 307: Transport Phenomena in Biological Systems course over the Fall 2025 semester at Duke University. The course introduces the modeling of complex biological systems through the principles of transport phenomena and biochemical kinetics, covering fundamental concepts such as the conservation of mass and momentum using differential and integral balances, the rheology of Newtonian and non-Newtonian fluids, steady and transient diffusion in reacting systems, dimensional analysis, and homogeneous versus heterogeneous reaction systems. It provides students with a rigorous foundation for analyzing and solving biomedical and biotechnological transport and reaction problems. We extend our heartfelt congratulations on this well-deserved honor and wish her continued success in her future endeavors. 🏆 #RandlesLab #DukeUniversity #BiomedicalEngineering #PrattEngineering #BME #TA #TeachingAssistant