Machine Learning Approaches for Hemodynamic Analysis of Non-Newtonian Blood Flow
Year: 2026
Type: book-chapter
Abstract: : Blood is a non-Newtonian, shear-thinning and viscoelastic fluid with complicated rheology, which significantly deviates from the assumptions of the Newtonian approach in small diameter arteries, stenotic regions and areas of low shear rate recirculation. Hence, accurate analysis of hemodynamics involves application of constitutive relations like Carreau or Casson along with Navier-Stokes equations, but purely numerical solution to these equations remains too costly for patient-specific and on-the-spot clinical applications. In this chapter we look at how machine learning approaches like artificial neural networks, convolutional neural networks, recurrent neural networks and physics-informed neural networks are used to accelerate computational hemodynamics, improve medical image processing and help diagnose cardiovascular diseases. First, the mathematical background of non-Newtonian blood flow modeling is presented, then the machine learning methods themselves and their applications in stenosis detection, flow prediction and patient-specific simulations are reviewed in detail. Two conceptual schemas demonstrate a generic approach to machine learning in computational hemodynamics and coupling between governing equations and learning algorithms. Constitutive models and machine learning approaches are summarized in comparative tables. Finally, the current limitations of machine learning in computational hemodynamics are discussed and promising directions for future research are outlined. Keywords: Non-Newtonian blood flow; Carreau model; Casson model; machine learning; physics-informed neural networks; computational fluid dynamics
Authors: Yokeshwari Ravi, Balaganesan Palanivelu
Language: English
Subfield: Statistical and Nonlinear Physics
Field: Physics and Astronomy
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7204527596