Adaptive Detection in Real-Time Gait Analysis through the Dynamic Gait Event Identifier

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Signal Processing

Background:

• Gait analysis is crucial for understanding human movement and diagnosing conditions.
• Real-time gait event detection is essential for developing responsive wearable systems.
• Existing methods often struggle with accuracy and efficiency in embedded applications.

Purpose of the Study:

• To introduce the Dynamic Gait Event Identifier (DGEI) for real-time gait event detection.
• To develop a method suitable for embedded system design and optimization.
• To accurately identify key gait events including heel strike (HS), toe-off (TO), walking start (WS), and walking pause (WP).

Main Methods:

• Software and hardware co-design approach for gait analysis.
• Real-time data analysis using first-order difference functions and sliding window techniques.
• Dynamic feature extraction including differential integration, weighted sleep time analysis, and adaptive thresholding on IMU signals.

Main Results:

• Achieved 97.82% accuracy for heel strike (HS) detection.
• Achieved 99.03% accuracy for toe-off (TO) detection.
• Demonstrated near-perfect alignment with human annotations (less than one frame difference in 99.2% of cases).

Conclusions:

• The DGEI provides a highly accurate and efficient solution for real-time gait event detection.
• The method's accuracy and suitability for embedded systems are validated by extensive experimental results.
• DGEI sets a new standard for gait analysis in wearable and embedded applications.