Deep-learning enabled smart insole system aiming for multifunctional foot-healthcare applications.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Artificial Intelligence in Healthcare

Background:

• Traditional smart insoles lack comprehensive foot health analysis.
• Manual feature extraction limits real-time plantar pressure mapping and gait analysis.
• Diverse user needs for advanced foot healthcare are unmet.

Purpose of the Study:

• To develop a deep learning-enabled smart insole system for comprehensive foot health monitoring.
• To enhance early prevention of foot diseases and provide personalized foot healthcare.
• To demonstrate the system's capability in exercise fatigue recognition and diabetic foot ulcer prevention.

Main Methods:

• A smart insole system integrating capacitive sensing, a portable circuit board, and deep learning algorithms.
• Utilizing statistical tools for long-term foot pressure data analysis and deep learning model training.
• Implementing a segmentation-assisted deep learning model for exercise-fatigue recognition.

Main Results:

• The capacitive insole accurately maps static and dynamic plantar pressure (>500 kPa) with high sensitivity.
• Deep learning algorithms uncover insights into plantar pressure patterns and foot issues.
• Exercise-fatigue recognition achieved 95% classification accuracy.

Conclusions:

• The deep learning smart insole system provides advanced, real-time foot health monitoring.
• The system supports early disease prevention, exercise injury avoidance, and diabetic foot ulcer prevention.
• This technology enhances quality of life through comprehensive foot healthcare solutions.