Derivative Based Gait Event Detection Algorithm Using Unfiltered Accelerometer Signals.

Area of Science:

• Biomedical Engineering
• Biomechanics
• Wearable Technology

Background:

• Gait event detection is crucial for analyzing walking patterns.
• Existing algorithms often require complex sensor fusion or are computationally intensive, limiting real-time applications.
• Current rules-based methods for real-time gait analysis can be complex and less user-friendly.

Purpose of the Study:

• To develop a novel, rules-based algorithm for accurate gait event detection using a single wearable sensor.
• To overcome the limitations of existing machine learning and multi-sensor approaches for real-time gait analysis.
• To enable the development of compact, user-friendly wearable systems for robust gait analysis.

Main Methods:

• Utilized a single accelerometer sensor placed on the foot.
• Developed a novel rules-based algorithm analyzing the derivative of the accelerometer signal.
• Implemented an optimizer and windowing approach for event detection.
• Tested the algorithm across various walking speeds and transitions.

Main Results:

• Achieved high performance in detecting heel-strike and toe-off events.
• Reported sensitivity of 94.32% and specificity of 94.70%.
• Demonstrated a low timing error of 6.52 ± 22.37 ms, even with speed changes.

Conclusions:

• The novel algorithm reliably detects key gait events using minimal sensor data.
• This approach is suitable for real-time applications, offering a simpler and more compact solution.
• Enables robust gait analysis in real-world settings for insights into gait quality and system control.