Modeling Arterial Blood Flow Using Physics-Informed Neural Networks

This paper focuses on the development of a computational model based on Physics-Informed Neural Networks (PINNs) to simulate arterial blood flow dynamics and its interaction with the arterial wall. PINNs combine the principles of deep learning with partial differential equations, such as the Navier-Stokes equations, to provide an accurate representation of biomechanical phenomena in the cardiovascular system. By incorporating boundary conditions, physical constraints, and specific data, this approach allows for robust simulations even in scenarios with incomplete or sparse data. The proposed framework was validated by predicting key hemodynamic parameters-such as pressure and velocity-across arterial networks. Results highlight its potential for rapid predictions post-training, making it a valuable tool for cardiovascular research and clinical applications. Future work will focus on extending the model to more complex physiological scenarios.