Towards blood flow in the virtual human: efficient self-coupling of HemeLB

Area of Science:

• Computational fluid dynamics
• Biomedical engineering
• Medical simulation

Background:

• The creation of a virtual human, a personalized digital replica of an individual, is a key goal in medical research.
• This digital replica aims to aid in patient diagnosis, treatment, and recovery.
• Simulating complex biological systems like blood flow presents significant computational challenges.

Purpose of the Study:

• To enhance the HemeLB lattice Boltzmann code for simulating 3D macroscopic blood flow at a full human scale.
• To improve the scalability and performance of blood flow simulations on large-scale computing systems.
• To integrate a self-coupling strategy for simulating both arterial and venous systems within the virtual human framework.

Main Methods:

• Utilized the HemeLB lattice Boltzmann code for blood flow simulation.
• Implemented advanced memory management and load balancing techniques for code optimization.
• Developed a self-coupling strategy to enable simultaneous simulation of arterial and venous vascular trees.

Main Results:

• Achieved near-linear scaling performance of the HemeLB code on hundreds of thousands of computer cores.
• Successfully simulated 3D macroscopic blood flow on a full human scale.
• Enabled simultaneous simulation of arterial and venous vascular trees using human-specific geometries.

Conclusions:

• Significant progress has been made in simulating human-scale blood flow, a crucial step towards the virtual human.
• The optimized HemeLB code demonstrates high scalability for complex physiological simulations.
• The integrated self-coupling strategy facilitates comprehensive vascular modeling for personalized medicine.