A Real-Time Health 4.0 Framework with Novel Feature Extraction and Classification for Brain-Controlled IoT-Enabled...

Area of Science:

• Neuroscience and Biomedical Engineering
• Human-Computer Interaction
• Internet of Things (IoT)

Background:

• Developing accurate motor imagery (MI) classification from electroencephalography (EEG) is crucial for brain-controlled environments.
• Existing methods, while improving, still face significant misclassification rates.
• Assisting disabled individuals in controlling smart environments (e.g., lighting, HVAC) requires robust and real-time brain-computer interfaces (BCIs).

Discussion:

• A novel framework combining regularized Riemannian mean (RRM) and linear SVM was proposed for four-class MI classification.
• A persistent decision engine (PDE) was introduced to significantly enhance MI classification accuracy (CA).
• The proposed methods were validated on two distinct four-class MI datasets, demonstrating improved performance over state-of-the-art techniques.

Key Insights:

• The RRM architecture achieved average CAs of 74.30% and 67.60% on two datasets.
• The PDE framework boosted average CAs to 92.25% and 82.54% on the same datasets.
• The PDE algorithm proves more reliable for four-class MI classification and is suitable for Brain-Controlled Environments (BCE).

Outlook:

• The study presents a low-complexity BCE architecture implemented in real-time on Raspberry Pi, suitable for wearable Health 4.0 devices.
• The findings suggest significant potential for advancing BCE research and applications.
• This work paves the way for more sophisticated and accessible assistive technologies.