Ambulatory Human Gait Phase Detection Using Wearable Inertial Sensors and Hidden Markov Model.

Area of Science:

• Biomechanics
• Medical Technology
• Data Science

Background:

• Gait analysis provides key kinematic and dynamic parameters.
• Abnormal gait phases are indicators for neurological diseases like Alzheimer's and Parkinson's.
• Current methods require further refinement for clinical application.

Purpose of the Study:

• To propose an inertial sensor-based gait analysis method.
• To accurately segment gait phases for disease diagnosis.
• To develop a reliable wearable device for gait data collection.

Main Methods:

• Utilized smoothed and filtered angular velocity signals as input.
• Employed a 15-dimensional temporal characteristic feature.
• Applied Hidden Markov Model (HMM) and parameter adaptive models for gait phase segmentation.

Main Results:

• The HMM and parameter adaptation model achieved high gait phase recognition rates.
• Segmentation results demonstrated consistency with ground truth.
• The wearable shoe-embedded device ensured stable, objective data collection in diverse environments.

Conclusions:

• The proposed method offers an accurate and reliable approach to gait phase segmentation.
• The wearable device facilitates real-world data acquisition for clinical applications.
• This technology has potential for early disease detection and monitoring.