Developing a method for quantifying hip joint angles and moments during walking using neural networks and wearables

Area of Science:

• Biomechanics
• Wearable Technology
• Artificial Intelligence

Background:

• Accurate quantification of hip angles and moments during gait is essential for diagnosing and treating hip pathologies.
• Wearable sensors combined with artificial neural networks (ANNs) offer a promising avenue for portable and cost-effective gait analysis.

Purpose of the Study:

• To develop and validate a Wearable-ANN approach for calculating hip joint angles and moments during walking.
• To assess the accuracy of this approach using data from healthy subjects.

Main Methods:

• Utilized a single shin-mounted inertial measurement unit (IMU) and a force-measuring insole for data collection.
• Developed a two-hidden-layer artificial neural network (ANN) with 5 nodes per layer.
• Trained and tested the ANN on data from 17 healthy subjects walking in the sagittal and frontal planes.

Main Results:

• The Wearable-ANN approach achieved an average root mean squared error (rRMSE) of 15%.
• The model demonstrated a high coefficient of determination (R ²) of 0.85 across various outputs, subjects, and training iterations.
• Performance was validated against a benchmark approach, showing comparable accuracy.

Conclusions:

• The developed Wearable-ANN system provides a reliable and accurate method for estimating hip joint angles and moments during gait.
• This technology has the potential to enhance diagnostic and treatment strategies for hip pathologies.
• The combination of wearable IMUs and ANNs represents a significant advancement in accessible biomechanical analysis.