An IoT-Enabled Stroke Rehabilitation System Based on Smart Wearable Armband and Machine Learning

Area of Science:

• Biomedical Engineering
• Rehabilitation Technology
• Internet of Things (IoT)

Background:

• Surface electromyography (sEMG) signals are crucial for hand function recovery training.
• Wearable devices for rehabilitation must prioritize user comfort and signal accuracy.
• Existing systems may face challenges with classification accuracy and user comfort.

Purpose of the Study:

• To develop an IoT-enabled stroke rehabilitation system integrating a smart wearable armband (SWA), machine learning (ML), and a 3D-printed robotic hand.
• To enhance user comfort and improve the accuracy of hand gesture recognition for stroke patients.
• To create a real-time, responsive robotic hand system for facilitating hand function recovery.

Main Methods:

• Developed a low-power IoT sensing device with textile electrodes in a wearable armband for sEMG signal measurement, pre-processing, and wireless transmission.
• Implemented an optimal feature set selection method and utilized ML algorithms for analyzing and discriminating hand movement features.
• Designed and implemented a 3D-printed five-finger robotic hand capable of mimicking user gestures in real-time.

Main Results:

• The smart wearable armband effectively measures, pre-processes, and transmits bio-potential signals with improved electrode distribution mitigating accuracy issues.
• Machine learning algorithms successfully identified all nine tested gestures with an average accuracy of up to 96.20%.
• The 3D-printed robotic hand accurately mimicked user gestures in real-time, demonstrating its potential as a rehabilitation tool.

Conclusions:

• The proposed IoT-enabled system, combining a comfortable SWA, advanced ML, and a dexterous robotic hand, shows significant promise for stroke rehabilitation.
• The system's high accuracy in gesture recognition and real-time robotic hand control facilitates effective hand function recovery training.
• This integrated approach offers a novel and effective training tool to aid patients in their post-stroke rehabilitation journey.