A scalable EEG-based spatial neglect detection system in augmented reality for stroke patients.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Computer Science

Background:

• Spatial neglect is a common post-stroke visuospatial attention disorder.
• Traditional pen-and-paper tests have limitations in clinical settings.
• AREEN, an AR-guided EEG system, was developed for neglect detection.

Purpose of the Study:

• To investigate the scalability of the AREEN system across individuals.
• To identify the most effective classification model for generalize neglect detection.
• To evaluate the system's ability to diagnose neglect in new patients.

Main Methods:

• Tested four classification models: logistic regression, linear discriminant analysis, random forest, and boosted tree.
• Employed a leave-one-participant-out cross-validation strategy.
• Assessed within-participant and within-group accuracies for neglect and non-neglect classification.

Main Results:

• The boosted tree model achieved the highest average within-participant accuracy: 76.0% for neglect and 68.2% for non-neglect.
• The boosted tree model demonstrated high within-group accuracy: 90.9% for neglect and 90.0% for non-neglect.
• Patients reported high satisfaction, comfort, and willingness to use the AREEN system.

Conclusions:

• The boosted tree model enables accurate spatial neglect identification, crucial for stroke rehabilitation planning.
• AREEN shows potential for real-time neglect detection with neurofeedback for future rehabilitation applications.
• The system's usability and patient acceptance are positive indicators for clinical adoption.