Classification of gait patterns in the time-frequency domain.

Area of Science:

• Biomechanics
• Wearable Technology
• Signal Processing

Background:

• Gait pattern analysis is crucial for understanding human locomotion and detecting abnormalities.
• Accurate classification of different gait activities, such as stair climbing and level walking, is essential for various applications, including healthcare and sports science.
• Existing methods for gait analysis can be complex or require specialized equipment.

Purpose of the Study:

• To develop and validate a reliable technique for classifying gait patterns during ascending stairs, descending stairs, and level walking.
• To assess the accuracy and effectiveness of using accelerometers for gait pattern classification.
• To provide a practical method for measuring gait during physical activity.

Main Methods:

• Utilized accelerometers placed on the shoulder in antero-posterior and vertical directions.
• Employed a two-step classification process involving wavelet coefficients.
• Applied direct spatial correlation of wavelet coefficients for segment separation.
• Used powers of wavelet coefficients from vertical and antero-posterior acceleration signals for feature classification.

Main Results:

• Achieved averaged absolute errors of 0.387 s for ascending stairs and 0.404 s for descending stairs compared to a reference system.
• Demonstrated high overall sensitivity and specificity for ascending stairs (98.79% and 99.52%) and descending stairs (97.35% and 99.62%).
• Successfully classified gait patterns during different physical activities.

Conclusions:

• The proposed technique reliably measures gait patterns during physical activity.
• Accelerometer-based gait classification offers a practical and accurate approach for analyzing locomotion.
• This method has potential applications in monitoring physical activity and clinical gait assessment.