Machine Learning based Human Gait Segmentation with Wearable Sensor Platform.

Area of Science:

• Biomechanics and Biomedical Engineering
• Machine Learning in Healthcare
• Wearable Sensor Technology

Background:

• Accurate gait segmentation is crucial for diagnosing and monitoring neurological and musculoskeletal disorders.
• Traditional methods often rely on complex laboratory equipment, limiting real-world applicability.
• Wearable sensors offer a promising, non-invasive approach for continuous gait analysis.

Purpose of the Study:

• To evaluate the effectiveness of supervised and unsupervised machine learning algorithms for gait segmentation.
• To utilize data from a multi-sensor wearable platform, including Inertial Measurement Units (IMUs) and plantar pressure insoles.
• To enable precise classification of gait phases and sub-phases using advanced data processing and machine learning techniques.

Main Methods:

• Collected gait data from 10 participants using a wireless wearable sensor platform (four IMUs and plantar pressure insoles).
• Preprocessed and annotated the anonymized gait data using Ranchos Los Amigos (RLA) gait nomenclature and a peak/valley detector.
• Implemented and compared unsupervised (K-Means clustering) and supervised (Support Vector Machine - SVM, Artificial Neural Network - ANN) machine learning algorithms for gait phase classification.

Main Results:

• Both supervised and unsupervised machine learning algorithms demonstrated capability in segmenting gait data.
• The implemented methods allowed for the classification of distinct gait phases and sub-phases.
• The study successfully applied advanced machine learning techniques to labeled datasets derived from wearable sensor data.

Conclusions:

• Wearable sensor technology combined with machine learning provides a viable solution for accurate gait segmentation.
• Supervised and unsupervised learning methods can effectively classify gait phases, offering potential for clinical applications.
• This approach facilitates objective and continuous gait monitoring outside of controlled laboratory settings.