Simultaneous Classification of Both Mental Workload and Stress Level Suitable for an Online Passive Brain-Computer...

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Passive Brain-Computer Interfaces (BCIs) typically focus on single mental state detection (e.g., mental workload).
• Real-world applications require simultaneous assessment of multiple user states, like cognitive and affective (stress/anxiety) states, for reliable BCI performance.
• Simultaneous state prediction is crucial for enhancing the practical effectiveness of online BCIs, especially in safety-critical domains.

Purpose of the Study:

• To investigate the feasibility of simultaneously classifying mental workload and stress levels within an online passive BCI.
• To compare subject-specific and cross-subject classification approaches.
• To evaluate the impact of transfer learning on cross-subject classification accuracy.

Main Methods:

• Utilized electroencephalography (EEG) signals for passive BCI.
• Implemented both subject-specific and cross-subject classification models.
• Applied transfer learning techniques to align data distributions for cross-subject analysis.
• Simulated an online analysis scenario across 18 participants.

Main Results:

• Achieved accuracies of 77.5 ± 6.9% for mental workload detection and 84.1 ± 5.9% for stress level detection using cross-subject classification with transfer learning.
• Demonstrated the effectiveness of transfer learning in improving cross-subject classification performance.
• Indicated that simultaneous classification is feasible and can yield reliable results.

Conclusions:

• Simultaneous classification of mental workload and stress is feasible in online passive BCIs.
• Transfer learning significantly enhances the performance of cross-subject classification for both mental workload and stress detection.
• These findings support the development of more robust and practical BCIs capable of monitoring multiple user states concurrently.