Undesirable properties of the dimensionless normalisation for spatio-temporal variables.

Area of Science:

• Biomechanics
• Gait Analysis
• Statistical Modeling

Background:

• Dimensionless normalization is commonly used for gait spatio-temporal parameters.
• This technique aims to remove the influence of body size, such as leg length.

Purpose of the Study:

• To investigate the potential undesirable properties of dimensionless normalization for gait data.
• To identify limitations and propose alternative methods for gait analysis.

Main Methods:

• Theoretical analysis of dimensionless normalization.
• Empirical data analysis of gait spatio-temporal parameters.
• Comparison of normalization techniques and statistical modeling approaches.

Main Results:

• Dimensionless normalization may not fully eliminate leg length correlation with gait parameters.
• The technique can introduce spurious correlations among spatio-temporal parameters, masking true relationships.
• Spurious correlations with external covariates are induced, complicating statistical modeling.

Conclusions:

• Alternative methods like residualization or using raw parameters with leg length as a regressor are proposed.
• Partial correlation is recommended for analyzing correlations between gait parameters.
• Careful consideration of analytical objectives is crucial when selecting normalization techniques for gait analysis.