Robust morris screening method (RMSM) for complex physiological models.

Area of Science:

• Computational modeling
• Biophysics
• Electrophysiology

Background:

• Complex, nonlinear models require efficient sensitivity analysis for parameter tuning.
• Classical Morris Screening Method (MSM) has limitations in reliability and reproducibility.
• Medical applications, like digital twins, necessitate robust tools for parameter assessment.

Purpose of the Study:

• Introduce a Robust Morris Screening Method (RMSM) for improved sensitivity analysis.
• Evaluate the reliability and stability of new RMSM indices (median absolute deviation and absolute median).
• Compare RMSM performance against classical MSM on a complex neuromuscular model.

Main Methods:

• Developed RMSM using novel statistical indices: absolute median (χ*) and median absolute deviation (ρ).
• Applied RMSM to a multi-scales neuromuscular electrophysiological model simulating HD-sEMG signals.
• Assessed index reliability across parameter space trajectories and compared with MSM.

Main Results:

• RMSM indices demonstrated normal distribution of elementary effects, unlike MSM.
• RMSM achieved stable parameter ranking with 20 trajectories, while MSM required 100.
• RMSM provided clearer distinction between influential and negligible model parameters.

Conclusions:

• RMSM offers a fast, reliable, and stable parameter ranking for complex models.
• The method enhances exploration of parameter influence, aiding future tuning and identification.
• RMSM surpasses classical MSM in efficiency and interpretability for sensitivity analysis.