Sensitivity Analysis of Upper Limb Musculoskeletal Models During Isometric and Isokinetic Tasks

Area of Science:

• Biomechanics
• Musculoskeletal modeling
• Computational physiology

Background:

• Sensitivity coefficients assess how parameter errors affect musculoskeletal model predictions.
• Previous lower limb studies lacked population diversity.
• Upper limb models require analysis with broader physiological variation.

Purpose of the Study:

• Perform sensitivity analysis on upper limb musculoskeletal models.
• Utilize a large synthetic dataset for enhanced physiological diversity.
• Compare methods for calculating sensitivity coefficients and their ability to account for measurement uncertainty.

Main Methods:

• Generated a dataset of 401 synthetic subjects with varied musculoskeletal parameters (e.g., maximum isometric force, optimal fiber length).
• Simulated isometric and isokinetic upper limb tasks using two musculoskeletal models.
• Calculated sensitivity coefficients using two-point, linear regression, and sensitivity function methods.

Main Results:

• Sensitivity functions demonstrated the best accounting for measurement uncertainty from medical imaging parameters.
• Muscle activations were most sensitive to optimal fiber length and maximum isometric force.
• Sensitivity levels varied depending on the specific muscle and task performed.

Conclusions:

• Large synthetic datasets effectively capture physiological diversity for musculoskeletal modeling.
• Optimal fiber length and maximum isometric force are critical parameters influencing upper limb muscle activation predictions.
• Sensitivity analysis results are muscle- and task-specific, informing model development and application.