A Novel Fabric Strain Sensor Array with Hybrid Deep Learning for Accurate Knee Movement Recognition

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Machine Learning

Background:

• Comprehensive knee joint monitoring is crucial for rehabilitation and sports science.
• Existing sensor systems often lack the ability to capture complex multi-axis kinematics and localized tissue deformation.
• A need exists for accurate, non-invasive, and lightweight solutions for knee movement analysis.

Purpose of the Study:

• To develop and evaluate a novel lightweight fabric strain sensor array for comprehensive knee joint monitoring.
• To assess the system's capability in capturing multi-axis knee kinematics and strain distribution.
• To investigate the effectiveness of a hybrid deep learning model for analyzing sensor data and classifying knee movements.

Main Methods:

• A two-layer fabric strain sensor array with eight sensing elements was designed.
• Ten subjects performed three activities (seated leg raise, standing, walking) while wearing the sensor array.
• A hybrid deep learning model, combining Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and Attention mechanisms, was employed for data analysis.

Main Results:

• The fabric strain sensor array effectively mapped strain distribution across the knee during movement.
• The hybrid deep learning model achieved 95% accuracy in recognizing fundamental knee movements.
• Channel attention analysis identified key sensors (2, 4, and 6) contributing significantly to classification performance.

Conclusions:

• The proposed fabric sensor array is a feasible and accurate system for monitoring knee joint movements.
• The hybrid deep learning model demonstrates strong capability in extracting spatial-temporal features for movement classification.
• This technology holds significant potential for applications in medical rehabilitation, sports science, and personalized healthcare.