Global Sensitivity Analysis of High-Dimensional Neuroscience Models: An Example of Neurovascular Coupling

Area of Science:

• Computational Biology
• Systems Biology
• Mathematical Modeling

Background:

• State-of-the-art cell models are increasingly complex, incorporating numerous parameters often derived from animal studies with uncertain relevance to human physiology.
• Parameter uncertainties in cell models propagate to output uncertainties, complicating the interpretation of model predictions and quantities of interest (QoIs).

Purpose of the Study:

• To develop and present a novel three-stage methodology for global sensitivity analysis (GSA) applicable to large and complex cell models.
• To apply this GSA methodology to a validated neurovascular coupling model with 160 uncertain parameters.

Main Methods:

• A three-stage methodology comprising screening (dimension reduction), surrogate modeling, and computation of Sobol' indices was employed.
• The methodology was applied to a numerical model of neurovascular coupling, analyzing three specific quantities of interest.

Main Results:

• The GSA methodology effectively apportioned output uncertainty to individual parameters for the analyzed neurovascular coupling model.
• Sensitivity analysis identified the relative influence of each parameter on three key quantities of interest: extracellular [Formula: see text], average volumetric flow rate, and minimum actin/myosin complex value.

Conclusions:

• The presented GSA approach provides a robust framework for analyzing complex cell models with numerous parameters.
• The findings offer insights into critical parameters influencing neurovascular coupling, highlighting areas for further physiological investigation and potential dysfunction identification.