Heart surgeons and cardiologists currently rely on simplified models of blood flow because the full physics of a beating heart is too complex for even powerful computers to simulate in a reasonable time. This project borrows mathematical tools from quantum physics—specifically, tensor networks—to compress the vast amount of data needed to model blood flow, turning an impossibly slow calculation into a fast, efficient one. The problem is that existing cardiovascular simulations are so computationally expensive that they cannot be used routinely for individual patients or for rapid testing of new treatments. By adapting tensor network methods to handle the moving walls and complex geometries of real arteries and heart chambers, the team aims to cut the time and cost of high-fidelity simulations by orders of magnitude. If successful, this would allow clinicians to run personalised blood-flow models for a patient in minutes rather than days, enabling tailored surgical planning for conditions such as aortic aneurysms or congenital heart defects. It would also accelerate fundamental research into cardiac biomechanics and the design of stents, valves, and other interventions. The work is primarily fundamental science—adapting quantum-inspired algorithms to a new domain—but with a clear path toward practical medical tools.
View original technical description
Mechanistic flow modelling is a critical tool in the study of cardiac pathophysiology, enabling detailed simulations of blood flow and pressure. However, its clinical and research utility is limited by the substantial computational cost associated with resolving complex cardiovascular dynamics. In quantum physics, tools such as tensor networks are used to reduce similarly difficult calculations into tractable problems. This project aims to adapt these tools from quantum physics to fluid simulation, investigating how they can be used to model cardiac flow with limited computational resources. Recent research has demonstrated that tensor networks can efficiently compress fluid velocity fields by exploiting the scale-locality of turbulence and the optimal low-rank representations achieved through truncated singular value decompositions. This compression not only significantly reduces storage requirements but also enables substantial computational speedups. By operating on the compressed representation, variational methods can be applied to simulate fluid dynamics with greatly improved efficiency. The project aims to extend this work to model biological flows by developing methods for treating the boundary conditions found in the cardiovascular domain (such as moving walls and anatomy-based geometries), exploring strategies for further reducing the computational cost of tensor network methods, and designing or adapting algorithms for implementation on quantum computers. By advancing these techniques, the project seeks to enable high-fidelity cardiovascular flow simulations that can be performed rapidly and cost-effectively. Such advances would facilitate personalised diagnostics and treatment planning, as well as accelerated research in cardiac biomechanics and intervention design.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know