Influence of multibody kinematic optimisation pipeline on marker residual errors

Area of Science:

• Biomechanics
• Musculoskeletal modeling
• Kinematics

Background:

• Residual errors are commonly used to assess musculoskeletal model fit to experimental data in multibody kinematic optimization (MKO).
• The validity of residual errors as a performance metric has been questioned due to potential non-causal relationships with computed joint angles.
• The influence of MKO pipeline parameters on residual errors remains largely unanalyzed.

Purpose of the Study:

• To investigate the effect of individual MKO pipeline steps on residual errors.
• To determine the existence of a causal relationship between residual errors and computed joint angles.

Main Methods:

• Systematic analysis of MKO pipeline parameters, including marker registration, segment scaling, marker weighting, and joint constraints.
• Quantification of residual errors and joint angle variations under different parameter settings.
• Statistical correlation analysis between residual errors and joint angle errors.

Main Results:

• Increases in residual errors were observed with marker registration errors (16.36 mm), segment scaling errors (14.84 mm), marker weighting variations (15.28 mm), and differing joint constraints (18.21 mm) compared to the baseline (13.84 mm).
• Significant variations in residual errors led to significant variations in computed joint angles.
• A positive correlation was found between increased residual errors and increased joint angle errors within the same MKO pipeline.

Conclusions:

• The MKO pipeline significantly influences residual errors, supporting a causal relationship between residual errors and joint angle accuracy.
• Findings challenge the sole reliance on residual errors as a goodness-of-fit metric, especially without artifact-free bone movement data.
• Results emphasize the need to carefully consider MKO pipeline parameters when interpreting residual errors for model validation.