Empirical Study on Human Movement Classification Using Insole Footwear Sensor System and Machine Learning.

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Human Movement Analysis

Background:

• Accurate human movement detection is crucial for various applications, including healthcare and sports.
• Existing methods often require specialized equipment or controlled environments.
• Developing unobtrusive, wearable systems for real-time movement monitoring is a significant challenge.

Purpose of the Study:

• To develop and validate a plantar pressure sensor system (P2S2) integrated into shoe insoles.
• To detect a comprehensive set of thirteen common human movements.
• To evaluate the efficacy of machine learning algorithms in classifying these movements based on pressure data.

Main Methods:

• Six Force Sensitive Resistor (FSR) sensors were embedded in shoe insoles to capture plantar pressure variations.
• Data from 34 adult participants performing thirteen distinct movements were collected.
• Principal Component Analysis (PCA) was used for data dimensionality reduction, followed by classification using k-NN, neural network, and Support-Vector Machine (SVM) algorithms.
• A four-fold cross-validation strategy was employed, ensuring subject independence during model training.

Main Results:

• The P2S2 system successfully captured distinct pressure signatures for thirteen human movements.
• Machine learning models, particularly SVM, demonstrated high performance in movement classification.
• An overall accuracy of 86% was achieved in predicting human movements using the integrated sensor system.

Conclusions:

• The P2S2 system provides a feasible and effective method for non-invasive human movement detection.
• The integration of plantar pressure sensing with machine learning offers a promising avenue for advanced human activity recognition.
• This technology has potential applications in gait analysis, fall detection, and personalized health monitoring.