On the calculation of sample entropy using continuous and discrete human gait data.

Area of Science:

• Biomechanics
• Biomedical Engineering
• Data Analysis

Background:

• Sample entropy (SE) is a reliable measure for time-series data, typically requiring over 500 discrete data points.
• Continuous data is often used when sufficient discrete data is unavailable, but its impact on SE is not well understood.
• The influence of sampling rate and input parameters (m, r) on SE for both continuous and discrete data requires investigation.

Purpose of the Study:

• To investigate the effect of using continuous versus discrete data on Sample entropy (SE).
• To determine the relative effects of sampling rate and input parameters (m, r) on SE for different data types.
• To provide recommendations for SE data collection and analysis in biomechanics.

Main Methods:

• Continuous lower-extremity joint angle data (480Hz) were collected from 11 subjects during walking.
• Data were downsampled to 240, 120, and 60Hz, and discrete range-of-motion (ROM) was calculated.
• SE was computed for both joint angles and ROM across various sampling rates and parameter (m, r) combinations.

Main Results:

• SE analysis revealed differential relationships between joint angles and ROM across lower extremity joints.
• SE of ROM showed no significant differences across sampling rates, indicating robustness.
• SE of joint angles decreased significantly from ankle to knee to hip.
• Continuous data exhibited greater sensitivity to changes in sampling rate and parameters (m, r) compared to discrete data.

Conclusions:

• Discrete data, especially ROM, is more robust and reliable for SE calculations than continuous data.
• Different data types (continuous vs. discrete) should not be directly compared for SE analysis.
• Utilizing discrete data is recommended for consistent and valid SE quantification in biomechanical studies.