Estimation of Fine-Grained Foot Strike Patterns with Wearable Smartwatch Devices.

Area of Science:

• Biomechanics
• Wearable Technology
• Machine Learning

Background:

• Foot striking (FS) patterns during locomotion can influence exercise benefits and injury risk.
• Accurate FS pattern recognition is crucial for optimizing gait and preventing injuries.

Purpose of the Study:

• To develop an intelligent system for recognizing subtle differences in foot striking patterns during walking and running.
• To achieve portable and non-invasive estimation of FS patterns using wearable smartwatch data.

Main Methods:

• A wearable system was developed to measure inertial hand movements using a smartwatch.
• Multivariate time series data from participants walking and running were captured.
• Deep neural network models, including 1D-CNN and GRUs, were employed for supervised learning classification.

Main Results:

• The proposed system achieved high and robust classification performance for FS patterns.
• Weighted-average F1 scores exceeding 90% were obtained, demonstrating system accuracy.
• The approach successfully correlated hand movements with distinct foot striking styles.

Conclusions:

• The developed intelligent system effectively recognizes foot striking patterns using wearable smartwatch data.
• This non-invasive method offers a practical approach to monitoring and potentially improving gait.
• Future applications could enhance exercise benefits and reduce injury risks through proper gait analysis.