More than a model: How digital twins are reshaping healthcare

This article is sponsored by OmniScale but was written by Cristin Merritt, Communications Chair of SC25, the International COnference for HIgh Performance Computing, Networking, Storage, and Analysis. 

Imagine your doctor holding up a tablet showing you two beating hearts. One is yours, monitored in real time by sensors tracking its rhythms. The other is also yours – a digital replica, built to test medications and procedures on your unique cardiovascular system before they ever reach your body. 

This isn’t science fiction, and it’s not tomorrow’s technology. These are digital twins – real-time, data-driven simulations already being used to plan surgeries, speed up drug development, and personalize medical treatments, and they’re being used today. 

Andrea Townsend-Nicholson, Ph.D., Vice Dean for Health in the Faculty of Life Sciences at University College London, is helping lead this transformation. A pioneer in biomedical applications of digital twins, her work bridges molecular biology and High-Performance Computing (HPC). She champions the use of powerful simulations that model systems from the molecular level up to entire organs, accelerating discovery, improving outcomes, and making science more sustainable and inclusive.  

Her efforts extend well beyond the lab. As a co-chair of The 3rd Digital Twins Workshop for High-Performance Computing at SC25, Townsend-Nicholson is helping to shape an international conversation about how digital twins, whether modeling human hearts or entire data centers, are changing the future of science, policy, and infrastructure. 

The digital twin moment

“Digital twins aren’t a futuristic concept anymore,” Townsend-Nicholson told us. “We’re already using them to model the human heart, simulate surgeries, and accelerate drug discovery. They let us ask and answer questions that used to be impossible without touching a patient or running a trial.” 

She speaks from experience. Her decades-long career in biomedical research has taken her across four continents, from cloning novel drug discovery targets in Sydney to decoding receptor protein interactions in London.. But her journey into digital twins began with a sense of frustration. As an experimental molecular biologist studying how cells communicate, she spent long months testing how a change in a single amino acid in the protein sequence can affect the protein’s function, making and testing these changes one by one in the lab. The process was slow, costly, and often showed an effect without a clear molecular explanation. 

Everything changed when she had the opportunity to collaborate with colleagues who were using HPC to model biological systems. 

“I started with simulations of proteins,” she explained. “Then I saw colleagues modeling blood flow, simulating digital vasculature, even animating a beating virtual heart. When we connected these separate simulations, it suddenly clicked. We weren’t just modeling parts anymore. We were building a virtual human.”  

That realization shifted her entire approach to science. What once took months of trial and error could now be simulated in hours, studied, and refined, before a single drop of blood was put into a test tube.  

“Digital twins didn’t just accelerate my research,” she says. “They expanded what was possible.” 

AI predicts, digital twins prove 

Artificial Intelligence (AI) continues to make headlines for its ability to automate tasks, identify patterns, and forecast outcomes. At the same time, digital twins are quietly reshaping how we understand and interact with complex systems. So while AI is general, digital twins are specific. And where AI makes broad predictions, digital twins focus on personalised modeling.  

As Townsend-Nicholson explains: “If the issue at hand closely mirrors patterns found widely within a population, AI is an excellent choice. But the more unique the case, the more critical it becomes to rely on detailed physics-based computational modeling.”  

Rather than competing, AI and digital twins often work best together. AI can detect trends, generate predictions, and guide decisions; digital twins can test those decisions, incorporating real-time data and simulating outcomes in a controlled virtual environment. This interplay opens up powerful new possibilities, particularly in domains where precision and personalization are essential. 

Unlike static models or historical analysis, digital twins are dynamic and responsive. They are continuously updated with sensor data, refined through experimentation, and used to generate insights that inform real-world actions. The connection between the simulation and its physical counterpart creates a living feedback loop; data flows in, decisions come out, and the model evolves alongside reality. 

The price of precision

For Townsend-Nicholson and her collaborators, the growing power of digital twins comes with a host of challenges, ethical, technical, and practical. Working with biomedical data at this level raises important questions: Who owns the data? How is model accuracy verified? Could heavy reliance on simulation influence medical decisions in unintended ways? 

“The technology needs to meet the person where they are,” she cautions. “A digital twin should reflect the patient’s needs, not just what the system assumes.” 

There’s also the question of scale. Creating digital twins that are responsive, predictive, and trustworthy requires significant computational power.  

“You might not need an exascale machine to run a patient’s twin every day,” she explains, “but you do need it to build and understand the model in the first place.”  

As digital twin adoption grows, there is a growing demand for fair and transparent access to high-performance computing resources. Researchers are asking not only how these tools are developed, but who gets to use them, and under what conditions. “The amount of energy needed to create and operate digital twins is substantial,” Townsend-Nicholson notes. “This raises challenging questions. Who gets to use this limited resource, and how do we prioritize that access?” 

A field that needs you 

“No one builds a digital twin alone,” Townsend-Nicholson says. “Every project requires collaboration across disciplines. From biologists, computer scientists, engineers, data governance experts – each brings a part of the picture, but none of us has all of it.”  

Digital twins represent more than just a technological breakthrough. They reflect a broader shift in how we approach problem-solving across fields. Already, they are being used to test treatments before prescriptions are written, to manage transportation systems in real time, and to model energy grids as they respond to shifting weather and demand. In climate science, researchers are building planetary-scale simulations like the Earth Systems Digital Twin to better understand atmospheric patterns, ocean dynamics, and future risks.  

The growing use of digital twins is creating new opportunities for people in every sector, not just those in traditional STEM fields. If your work involves systems, interactions, or data, there’s a good chance it could benefit from simulation. This is not a closed field; it is an open frontier where diverse perspectives and unconventional collaborations are essential. 

“People shouldn’t hesitate to step outside their comfort zones and explore areas slightly different from their own. I took that leap myself, and have never regretted it.” What’s more, she says, is that she has received tremendous support along the way. “No single person ever has all the important pieces,” Townsend-Nicholson encourages. “Digital twin projects thrive on that cross-pollination. Biologists, computer scientists, engineers, and data experts all bring a part of the picture, and together it becomes something transformative.” 

That spirit of collaboration is what drives the Digital Twins for HPC Workshop at SC25. Now in its third year, the workshop brings together researchers and practitioners from biosciences, energy, climate, aerospace, and infrastructure to collaborate and learn. Through shared methods, case studies, and open discussion, the event highlights not only what’s happening now, but what’s possible next.  

Digital twins rely on powerful computing systems, but their true potential lies in the creativity and curiosity of the people building them.  

Digital twins at SC25

Whether you’re designing infrastructure, advancing personalized medicine, or exploring how AI and simulation can work together, there’s a place for you in the digital twin conversation. 

The Digital Twins for HPC Workshop at SC25 presents a unique opportunity to connect with researchers, developers, and decision-makers from diverse disciplines. Through technical sessions, lightning talks, and real-world case studies, the workshop showcases how digital twins are being applied today and what’s coming next. 

As part of SC25, the world’s premier conference on high-performance computing, AI, and scientific innovation, the workshop invites you to be part of the collaborations shaping the future of simulation. 

Join us at SC25 in St. Louis, November 16–21, 2025, and be part of the collaborations that are pushing the boundaries of what simulations can model, solve, and achieve.