Prediction of Lower Extremity Multi-Joint Angles during Overground Walking by Using a Single IMU with a Low Frequency...

Area of Science:

• Biomechanics
• Wearable Technology
• Machine Learning

Background:

• Joint angle analysis is crucial for gait assessment, injury risk, and rehabilitation.
• Current inertial measurement unit (IMU) sensor methods for gait analysis face challenges in daily use due to multiple sensors and high battery consumption.
• Existing high-frequency IMU systems are inconvenient for long-term, real-world gait monitoring.

Purpose of the Study:

• To develop and validate a novel multi-joint angle estimation method using a single, low-frequency IMU sensor.
• To overcome the limitations of current IMU-based gait analysis systems for practical, long-term application.
• To assess the accuracy of the proposed method compared to traditional motion capture systems.

Main Methods:

• A long short-term memory (LSTM) recurrent neural network was employed for multi-joint angle estimation.
• Data was collected from a single IMU sensor attached to the lateral shank of 30 healthy young individuals during overground walking.
• Low-frequency (23 Hz) sensor data was utilized to reduce power consumption and enhance usability.

Main Results:

• The proposed method achieved good accuracy, comparable to studies using high-frequency IMU sensors.
• The coefficient of determination (R2) exceeded 0.74, with root mean square error (RMSE) below 7° and normalized RMSE (NRMSE) below 9.87% against motion capture data.
• The knee joint exhibited the highest estimation accuracy (best NRMSE and R2) among the evaluated hip, knee, and ankle joints.

Conclusions:

• A single, low-frequency IMU sensor combined with an LSTM network can accurately estimate multi-joint angles during gait.
• This approach offers a practical and potentially more convenient alternative for gait analysis in daily life and rehabilitation settings.
• The findings suggest a promising direction for developing user-friendly and efficient wearable gait monitoring systems.