On the choice of multiscale entropy algorithm for quantification of complexity in gait data

Area of Science:

• Biomechanics
• Nonlinear dynamics
• Signal processing

Background:

• Stride-to-stride variability is crucial for gait analysis and stability.
• Multiscale entropy (MSE) analysis offers a method to quantify complexity in time series data.
• Selecting an appropriate MSE algorithm is essential for reliable gait complexity assessment.

Purpose of the Study:

• To identify the most suitable multiscale entropy (MSE) algorithm for assessing complexity in stride-to-stride interval time series.
• To compare the performance of five different MSE algorithms using simulated data.
• To evaluate the chosen algorithms' efficacy in differentiating gait complexity between overground and treadmill walking.

Main Methods:

• Five MSE algorithms were tested on white noise, pink noise, and sine wave signals.
• Algorithm performance was evaluated based on complexity differentiation, sensitivity, and parameter consistency.
• The most appropriate algorithms (MSE and RCMSE) were applied to stride interval data from 14 healthy subjects during overground and treadmill walking.

Main Results:

• Multiscale entropy (MSE) and refined composite multiscale entropy (RCMSE) demonstrated superior ability to differentiate signal complexity.
• Both MSE and RCMSE exhibited acceptable sensitivity and good parameter consistency for gait data.
• Neither MSE nor RCMSE could significantly differentiate stride-to-stride interval complexity between overground and treadmill walking conditions.

Conclusions:

• Multiscale entropy (MSE) and refined composite multiscale entropy (RCMSE) are recommended for quantifying complexity in stride-to-stride interval time series.
• Further research may be needed to refine methods for detecting subtle differences in gait complexity between walking conditions.
• The findings provide guidance for selecting appropriate complexity analysis tools in gait research.