Mental fatigue prediction during eye-typing.

Area of Science:

• Neurology
• Human-Computer Interaction
• Biomedical Engineering

Background:

• Mental fatigue is a prevalent issue in neurological disorders, lacking objective, continuous measurement tools.
• Eye-typing systems are vital for communication in individuals with severe neurological impairments.
• Current fatigue assessment relies on subjective, intermittent methods.

Purpose of the Study:

• To develop and validate a continuous, objective model for assessing mental fatigue during eye-typing.
• To explore the utility of eye-tracking metrics and task difficulty in fatigue modeling.

Main Methods:

• An eye-typing experiment was conducted with 18 healthy participants, collecting eye-tracking data (blink frequency, eye height, pupil diameter) and typing performance.
• Subjective mental fatigue and effort ratings were gathered using a six-point Likert scale.
• Random forest regression was employed to model mental fatigue using eye-tracking features and task difficulty.

Main Results:

• The developed model predicted subjective mental fatigue ratings with 22% lower mean absolute error compared to simulations.
• Incorporating task difficulty as a feature increased the model's explained variance by 9%.
• Eye-tracking features (blink frequency, eye height, pupil diameter) proved effective in predicting fatigue.

Conclusions:

• Objective, continuous measurement of mental fatigue during eye-typing is feasible using eye-tracking data.
• Task difficulty significantly influences mental fatigue and should be considered in fatigue models.
• This approach offers a non-intrusive method to monitor fatigue in individuals using assistive communication technologies.