An Integrated Electroencephalography and Eye-Tracking Analysis Using eXtreme Gradient Boosting for Mental Workload...

Area of Science:

• Neuroscience and Biomedical Engineering
• Cognitive Science in Medicine
• Surgical Education Technology

Background:

• Traditional mental workload assessment relies on subjective self-report scales, introducing bias.
• Mental workload is complex and varies across surgical tasks, necessitating objective evaluation methods.
• Identifying key contributing factors to mental workload is crucial for optimizing surgical procedures and training.

Purpose of the Study:

• To develop and validate machine learning models for predicting mental workload during surgical tasks.
• To integrate electroencephalogram (EEG) and eye-tracking data for enhanced workload prediction.
• To identify critical features influencing mental workload in surgical simulations.

Main Methods:

• Utilized EEG and eye-tracking data from 26 participants performing simulated surgical tasks (da Vinci simulator, FLS program).
• Developed an eXtreme Gradient Boosting (XGBoost) model for mental workload evaluation.
• Analyzed features including pupil diameter, task complexity, temporal lobe functional connectivity, and eye movement trajectories.

Main Results:

• XGBoost models achieved high predictive performance (R²: 0.81-0.83) across simulated surgical tasks.
• Key predictors included pupil diameter, task complexity, temporal lobe connectivity, and eye movement patterns.
• Integrating EEG and eye-tracking data significantly improved model performance (p < 0.05), except for the Pattern Cut task.

Conclusions:

• Machine learning models integrating multimodal data show strong potential for objective mental workload prediction in surgery.
• This approach can inform surgical task design and personalize surgical training programs.
• Further research may refine models for specific tasks and explore additional neurophysiological correlates of cognitive load.