Functional Brain States Measure Mentor-Trainee Trust during Robot-Assisted Surgery.

Area of Science:

• Neurosurgery
• Human-Robot Interaction
• Machine Learning

Background:

• Mutual trust is crucial in surgical teams, particularly in robot-assisted surgery (RAS), where complex human-robot interfaces increase relational dynamics.
• Assessing mentor-trainee trust in RAS is challenging, relying heavily on subjective evaluations.
• Mentor-trainee trust involves the mentor's assessment of the trainee's performance and their readiness to continue the procedure.

Purpose of the Study:

• To propose and validate a novel, objective method for assessing mentor-trainee trust during RAS.
• To leverage electroencephalography (EEG) to analyze mentor brain activity while observing trainees.
• To identify key brain activity features indicative of trust using machine learning.

Main Methods:

• Monitored EEG activity of mentor surgeons observing surgical trainees performing procedures.
• Quantified mentor brain activity using functional and cognitive brain state features.
• Employed machine learning classification to distinguish trustworthy from concerning trainee performances based on EEG data.

Main Results:

• Functional brain features alone were sufficient for classifying trust during simple surgical tasks.
• For complex tasks, cognitive features offered marginal accuracy improvements, but functional brain state features remained the primary drivers of classification performance.
• EEG patterns effectively differentiated mentor perceptions of trainee trustworthiness.

Conclusions:

• Functional brain network interactions contain valuable information for objective, trainee-specific mentorship in RAS.
• This objective trust assessment method can form the basis for automation in human-robot shared control environments.
• The findings pave the way for more data-driven and effective surgical training and collaboration.