Gait Cycle Validation and Segmentation Using Inertial Sensors.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Gait Analysis

Background:

• Accurate gait event detection is crucial for assessing mobility and diagnosing neurological conditions.
• Existing methods often lack real-time processing capabilities or robust performance across diverse populations.
• Inertial sensors offer a promising, non-invasive approach for continuous gait monitoring.

Purpose of the Study:

• To develop and validate a novel algorithm for real-time gait cycle segmentation and event detection using foot-mounted inertial sensors.
• To compare the algorithm's performance against existing methods in control, Parkinson disease, and geriatric populations.
• To assess the algorithm's effectiveness under both fixed and variable sampling rates.

Main Methods:

• Utilized physical models of sensor data to distinguish stationary and moving segments.
• Developed a sparsity-assisted wavelet denoising (SAWD) routine for generating gait cycle templates from gyroscope data.
• Employed root-mean-square error for gait cycle validation and local minima detection for gait event identification (midstance, toe-off, heel-strike).

Main Results:

• The proposed algorithm achieved an average F1 score of 87.78% across all participant groups at a fixed sampling rate.
• Demonstrated superior performance in Parkinson disease participants, achieving an average F1 score of 92.44% with a variable sampling rate.
• Successfully validated and segmented gait cycles in real-time, identifying key gait events.

Conclusions:

• The developed algorithm provides an accurate and efficient method for real-time gait event detection using inertial sensors.
• The SAWD technique offers robust performance, particularly beneficial for analyzing gait in individuals with Parkinson disease.
• This technology has significant potential for clinical applications in gait monitoring and rehabilitation.