A Novel Template-Matching Method for Extracting Gait Cycles from Underfoot Pressure Data.

Area of Science:

• Biomechanics
• Wearable Technology
• Data Science

Background:

• Accurate gait cycle isolation from underfoot pressure data is essential for human movement analysis.
• Current methods often demand costly equipment or extensive manual labor, limiting their practical application.
• There is a need for efficient and accessible tools for gait cycle detection.

Purpose of the Study:

• To introduce a generalizable, low-resource algorithm for parsing gait cycles from wearable underfoot pressure sensor data.
• To evaluate the algorithm's accuracy and processing time compared to manual methods and threshold-based approaches.
• To enable scalable and automated gait analysis in diverse walking environments.

Main Methods:

• Development of a novel algorithm for gait cycle parsing from underfoot pressure sensor data.
• Validation against a ground-truth dataset manually marked by an expert.
• Comparison of algorithm performance (accuracy, processing time) against manual parsing and threshold-based methods across various terrains.

Main Results:

• The algorithm demonstrated comparable accuracy to expert manual marking but with drastically reduced processing time (41 seconds vs. 29 minutes for 577 steps).
• The proposed method yielded only one false negative, significantly outperforming manual parsers with 6-33 errors.
• Threshold-based parsing methods were found to be less accurate, producing a high number of false positives (49-362).

Conclusions:

• The developed algorithm offers an efficient and accurate solution for gait cycle detection using wearable pressure sensors.
• This low-resource method enhances computational efficiency, making automated gait analysis more feasible for real-world applications.
• The findings support the expansion of automated gait analysis beyond controlled laboratory settings into practical, everyday environments.