Sensitivity Analysis of a Hydraulic Network Model of Glymphatic Flow

Flow of cerebrospinal fluid and extracellular fluid in the brain interior has significant implications for waste clearance and drug and nutrient delivery. However, limited in vivo access to the brain interior leaves gaping holes in human understanding and has led to debate on the nature of these flows. We developed a lumped-parameter network model representing the glymphatic system, based on available experimental measurements. Of the 23 parameters in the model, 11 are difficult to measure directly and have considerable uncertainty associated with them. We performed a sensitivity analysis using a Monte Carlo approach to determine how the model parameters affect quantities of interest, including total model hydraulic resistance, flow distribution, and parenchymal perfusion. The sensitivity analysis revealed that total model resistance is most sensitive to the permeability of penetrating perivascular spaces, which can motivate future experimental efforts. The input parameter space was reduced based on the range of parameters that require an unrealistically large driving pressure. Model features that are known or hypothesized to change with arousal state were varied to determine which parameters have the greatest potential to regulate glymphatic flow.