Adaptive gait segmentation algorithm for walking bout detection using tri-axial accelerometers.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Human Movement Analysis

Background:

• Gait analysis in daily life aids understanding of recovery and disease progression.
• Accurate identification of gait activity bouts is crucial for home-based monitoring.
• Existing methods often require subject-specific calibration, limiting practicality.

Purpose of the Study:

• To present a novel, fully automated method for detecting gait activity bouts using shank-mounted accelerometer data.
• To develop an adaptive thresholding algorithm that eliminates the need for subject-specific parameters.
• To validate the algorithm's performance against manual annotations and stopwatch timings.

Main Methods:

• Utilized shank-mounted accelerometers to record gait data from healthy individuals and a Parkinson's disease (PD) patient.
• Developed an automated algorithm with adaptive thresholding to identify gait bout onset and offset.
• Compared algorithm-derived bout times and durations with experimentally recorded stopwatch times and manual annotations.

Main Results:

• High agreement (ICC r > 0.97) between the algorithm and manual annotations for healthy subjects across various walking speeds.
• Moderate agreement (ICC r = 0.663) for a single PD subject.
• Low mean absolute errors for bout onset (0.91-1.17s), offset (0.80-2.41s), and duration (1.27-3.67s) detection.

Conclusions:

• The proposed automated gait bout detection method is reliable and accurate for healthy individuals.
• The adaptive thresholding approach offers a practical solution for home-based gait monitoring.
• Further validation in diverse patient populations is warranted for broader clinical application.