Ankle Angle Prediction Using a Footwear Pressure Sensor and a Machine Learning Technique.

Area of Science:

• Biomechanics
• Wearable Technology
• Occupational Health

Background:

• Ankle injuries elevate the risk of lower extremity joint damage and workplace impairments.
• Predicting and preventing these injuries requires accurate monitoring of biomechanical factors during common occupational tasks.

Purpose of the Study:

• To develop and validate a footwear-integrated pressure sensor system for predicting ankle angles.
• To utilize machine learning for accurate angle prediction during dynamic occupational movements.

Main Methods:

• A flexible footwear sensor incorporating six force sensing resistors (FSRs), a microcontroller, and Bluetooth LE was designed.
• Twenty-six subjects performed squat and stoop lifting motions, with ankle angles captured by the sensor system.
• The k-nearest neighbor (kNN) algorithm modeled ankle angle prediction, validated against an inertial measurement unit (IMU) system.

Main Results:

• The footwear sensor system achieved over 93% accuracy in predicting ankle angles during squat motions.
• An accuracy of 87% was recorded for stoop motions, demonstrating reliable performance.
• The system successfully correlated sensor data with actual ankle joint kinematics.

Conclusions:

• The developed plantar pressure sensor system is a viable and accurate tool for predicting ankle angles.
• This technology offers a promising approach for real-time injury prevention in occupational settings.
• Monitoring ankle kinematics during lifting tasks can mitigate risks associated with lower extremity injuries.