Real-time Mental State Recognition using a Wearable EEG.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Signal Processing

Background:

• Advancements in low-cost electroencephalography (EEG) systems enable non-medical applications.
• Existing real-time mental state assessment algorithms primarily rely on medical-grade EEG equipment.
• Consumer-grade wearable EEG devices present an opportunity for accessible mental state monitoring.

Purpose of the Study:

• To evaluate an approach for real-time focus and relax state detection using a consumer-grade wearable EEG headband.
• To introduce and assess novel entropy-based measures for EEG signal analysis.
• To develop and validate classifiers for mental state recognition.

Main Methods:

• Computed one naive and four entropy-based measures using relative EEG band powers.
• Developed classifiers based on probability distribution estimation for focus and relax states.
• Evaluated the performance of the developed approach in a user study.

Main Results:

• Tsallis entropy measure showed the best performance for focus state detection.
• Renyi measure demonstrated superior performance for relax state detection.
• Achieved high accuracy with sensitivities of 82.0% (Focus) and 80.4% (Relax), and specificities of 82.8% (Focus) and 80.8% (Relax).

Conclusions:

• Wearable EEG systems are viable for real-time mental state recognition.
• Entropy-based measures offer effective methods for analyzing EEG signals from consumer-grade devices.
• The developed approach successfully distinguishes between focus and relax states using accessible technology.