Physiology-based parameterization of human blood steady shear rheology via machine learning: a hemostatistics contribution

Abstract

Hemorheology is the study of blood flow and the mechanical stresses and kinematics involved. The Casson constitutive equation is a popular and simple model used to describe the steady shear rheology of blood, with only two parameters that specify an infinite shear viscosity and a yield stress that depend on blood physiology. Previous literature has identified hematocrit and fibrinogen concentration as the two most important physiological factors that affect blood flow, but previous parameterizations of the Casson model may not be reliable due to the use of non-standardized data sets. This study uses machine learning and the largest standardized dataset to improve the parameterization of the Casson model with respect to hematocrit and fibrinogen concentration for healthy individuals. The study also employs machine learning to identify a potential additional factor, the mean corpuscular hemoglobin (MCH), that may affect blood rheology. The proposed approach demonstrates the potential for machine learning to improve the connection between physiology and blood rheology with possible implications in cardiovascular diagnostics.