Modeling Arterial Blood Flow Using Physics-Informed Neural Networks.

Area of Science:

• Biomedical Engineering
• Computational Fluid Dynamics
• Artificial Intelligence in Medicine

Background:

• Cardiovascular diseases necessitate accurate simulation of blood flow dynamics.
• Traditional methods struggle with complex arterial geometries and incomplete data.
• Physics-Informed Neural Networks (PINNs) offer a novel approach by integrating physical laws with deep learning.

Purpose of the Study:

• To develop and validate a computational model using PINNs for simulating arterial blood flow and wall interactions.
• To accurately represent biomechanical phenomena in the cardiovascular system.
• To enable robust simulations even with sparse or incomplete datasets.

Main Methods:

• Development of a computational framework leveraging Physics-Informed Neural Networks (PINNs).
• Integration of the Navier-Stokes equations and relevant boundary conditions within the PINN architecture.
• Incorporation of physical constraints and specific datasets for model training and validation.

Main Results:

• The PINN-based model successfully simulated arterial blood flow dynamics and wall interactions.
• Accurate prediction of key hemodynamic parameters, including pressure and velocity, across arterial networks.
• Demonstrated potential for rapid, accurate predictions post-training.

Conclusions:

• The developed PINN model provides a powerful and efficient tool for cardiovascular research.
• This approach enhances the simulation of biomechanical phenomena in the cardiovascular system.
• The framework shows promise for clinical applications and future extensions to complex physiological scenarios.