Personalized Parameter Setting in Musculoskeletal Models Through Multitrajectory Optimization.

Area of Science:

• Biomechanics
• Computational Modeling
• Human Movement Analysis

Background:

• Personalized musculoskeletal models are crucial for understanding muscle and joint mechanics.
• Parameter non-identifiability in these models hinders accuracy and reliability due to compensating parameters.
• Existing single-task optimization methods face limitations in precise parameter estimation.

Purpose of the Study:

• To introduce and validate a multitrajectory optimization framework integrated with subject-specific modeling.
• To address the challenge of parameter non-identifiability in personalized musculoskeletal models.
• To enhance model accuracy and robustness through diverse movement task incorporation.

Main Methods:

• Developed a dual-stage optimization process: global search with Particle Swarm Optimization (PSO) and local refinement with Pattern Search.
• Integrated diverse movement tasks (biceps curl variations) to constrain the parameter space.
• Applied the framework to subject-specific modeling for parameter estimation.

Main Results:

• Achieved a 97.9% reduction in optimization convergence error and a 99.2% reduction in validation error on an unseen task.
• Demonstrated improved parameter estimation accuracy and model robustness under specific conditions.
• Showcased the generalizability of the framework across tested movement conditions.

Conclusions:

• The multitrajectory optimization framework significantly enhances parameter estimation accuracy in musculoskeletal models.
• This approach offers a promising solution for improving the reliability and precision of subject-specific biomechanical models.
• Preliminary insights suggest potential applications in clinical rehabilitation, sports science, and ergonomic design.