Uncertainty quantification and sensitivity analysis of neuron models with ion concentration dynamics

Area of Science:

• Computational Neuroscience
• Mathematical Biology
• Biophysics

Background:

• Neuron models with ion concentration dynamics present significant challenges for uncertainty quantification (UQ) and global sensitivity analysis (GSA) due to computational cost and complex dynamics.
• Existing methods struggle with parameters affecting resting states and the interplay of fast (electrical potentials) and slow (ion concentrations) dynamics.

Purpose of the Study:

• To develop and apply computationally efficient UQ and GSA methods to a detailed neuron model (edNEG) that includes ion concentration dynamics.
• To address challenges in UQ/GSA for complex neuron models, including computational expense and parameter influence on resting state and dynamics.
• To identify key parameters driving neuron model behavior under physiological and pathological conditions.

Main Methods:

• Utilized a variance-based GSA approach to pinpoint influential input parameters.
• Employed surrogate modeling and efficient numerical integration to reduce computational burden.
• Developed a strategy to isolate parameters affecting the neuron's resting state.
• Analyzed the electrodiffusive neuron-extracellular-glia (edNEG) model, incorporating six compartments and dynamics of key ions (Na+, K+, Ca2+, Cl-) and volume.

Main Results:

• Successfully quantified uncertainty and identified critical parameters in the edNEG model.
• Demonstrated the effectiveness of surrogate modeling for computationally intensive UQ and GSA.
• Characterized the influence of uncertain parameters on both rapid spiking dynamics and slower ion concentration changes.
• Provided insights into model behavior under physiological and pathological conditions.

Conclusions:

• The developed UQ and GSA methodology is computationally efficient and applicable to complex neuron models with ion dynamics.
• The study offers practical guidelines for future research in computational neuroscience, enhancing the reliability and interpretability of neuron models.
• This work contributes to a deeper understanding of how parameter uncertainty affects neuronal function and dysfunction.