Characterizing Bodyweight-Supported Treadmill Walking on Land and Underwater Using Foot-Worn Inertial Measurement...

Area of Science:

• Biomechanics
• Rehabilitation Engineering
• Machine Learning

Background:

• Gait rehabilitation often uses bodyweight unloading, like overhead support or underwater buoyancy.
• Inertial Measurement Unit (IMU) sensors offer a wireless, cost-effective way to measure body motion for gait analysis.
• Automated gait event detection from IMU data is challenging, requiring precise parameter tuning.

Purpose of the Study:

• To evaluate machine learning for detecting gait events using IMU sensors on the foot.
• To assess this method during bodyweight-supported treadmill walking, both on land and underwater.
• To determine the accuracy of IMU-based gait event detection with machine learning.

Main Methods:

• Twelve healthy subjects walked on a treadmill with overhead bodyweight support; three subjects walked underwater.
• Inertial Measurement Unit (IMU) sensors were placed on the foot, with motion capture and ground reaction force data collected on land.
• Wireless foot pressure insoles recorded IMU data underwater; random forest classification was used for gait event detection.

Main Results:

• High accuracy (95-96%) was achieved for gait event detection during on-land, bodyweight-supported treadmill walking.
• Underwater gait event detection required specific training data due to altered biomechanics.
• Single-axis IMU data and machine learning effectively identified gait events in both conditions.

Conclusions:

• Machine learning classification using foot-worn IMU data enables accurate gait event detection.
• Automated detection methods are crucial for advancing gait rehabilitation technologies.
• This approach offers a robust alternative to traditional gait analysis methods.