A DIAMETER AND HEMODYNAMIC REGULATION MODEL UTILIZING AN ARTERIOLAR NETWORK STRUCTURAL MODEL AND EXPERIMENT-BASED MACHINE LEARNING

Arteriolar regulation is central to maintaining adequate tissue perfusion, and its disruption can lead to adverse health outcomes. However, mechanisms governing vessel responses to complex environmental conditions within a network context remain uncertain. This study presents computational simulations of arteriolar network responses to various combinations and intensities of five physiologically significant stimuli: metabolic (via adenosine), adrenergic (via norepinephrine), oxygen, myogenic (intralumenal pressure), and blood flow/shear stress. Individual vessel diameter adaptations are predicted using machine learning models trained on extensive experimental data of arteriolar responses to stimuli combinations in rat skeletal muscle. After evaluating multiple single vessel modeling pipelines, a high-performing machine learning vessel model (R ² > 0.9) was applied to an initially static (i.e., non-regulating) network to obtain a novel steady-state network regulation model. The initial static arteriolar network was obtained via a previously described experimentally validated constrained constructive optimization (CCO) approach. Regulated network characteristics are analyzed based on diameter distribution and over a range of hemodynamic properties (volumetric blood flow, red blood cell flow, tube and discharge hematocrit) relevant to tissue health. These simulations support a range of research objectives, allowing user adjustments to network properties (e.g., vessel density), tissue shape/size, and environmental conditions (i.e., varying intensities and combinations of the five vasoactive stimuli). The methodology in this study promotes accessible modeling and holds application in disease research or therapy developments, as well as permitting further theoretical development such as time-dependent network regulation.