Related Experiment Video
Updated: Sep 21, 2025
12:51
Evaluation of Commercial-Off-The-Shelf Wrist Wearables to Estimate Stress on Students
Published on: June 16, 2018
7.6K
Automated Classification of Cognitive Workload Levels Based on Psychophysiological and Behavioural Variables of
Monika Kaczorowska1, Małgorzata Plechawska-Wójcik1, Mikhail Tokovarov1
1Department of Computer Science, Lublin University of Technology, 20-618 Lublin, Poland.
Brain Sciences
|May 28, 2022
Summary
This study effectively classifies cognitive workload levels using ex-Gaussian eye-tracking and cognitive data. Machine learning achieved nearly 96% accuracy, highlighting key features for workload assessment.
Area of Science:
- Cognitive Science
- Human-Computer Interaction
- Machine Learning
Background:
- Cognitive workload assessment is crucial for performance and safety.
- Traditional methods often lack objective, real-time measures.
- Eye-tracking and cognitive tests offer potential for nuanced workload evaluation.
Purpose of the Study:
- To apply ex-Gaussian parameters from eye-tracking and cognitive measures for classifying cognitive workload levels.
- To develop and utilize a computerized Digit Symbol Substitution Test (DSST) for data collection.
- To evaluate the effectiveness of ex-Gaussian modeling and machine learning in workload classification.
Main Methods:
- Collected eye-tracking data (saccades, fixations, blinks) and DSST performance metrics (response time, correct responses).
- Applied ex-Gaussian modeling to analyze the distribution of collected data.
- Employed independent classification using classical machine learning methods.
Main Results:
- Ex-Gaussian modeling effectively identified group dissimilarities.
- Achieved an overall classification accuracy of nearly 96% for cognitive workload levels.
- Identified key features influencing workload classification using logistic regression.
Conclusions:
- Ex-Gaussian modeling combined with eye-tracking and cognitive data provides a robust method for workload classification.
- Machine learning models can accurately predict cognitive workload with high precision.
- Feature importance analysis offers insights into the underlying cognitive processes during tasks.
