Presentation
Predicting Cognitive Workload from Pupillometry: A Machine Learning Approach
SessionPoster Session 1
DescriptionMonitoring and controlling cognitive workload are critical for maintaining smooth performance in complex visuospatial tasks requiring high mental effort. This study utilizes pupillometry with machine learning models to predict cognitive workload for a Lego building task. Twenty university students were recruited to participate in the complex Lego building task while pupil diameter data were collected using an eye tracker. After the data were preprocessed and normalized, the K-means clustering method classified baseline-corrected pupil data into moderate and high workload levels. Random forest, XGBoost, and Long Short-Term Memory (LSTM) were trained to predict the workload levels. Results showed moderate prediction accuracy, with LSTM providing better accuracy than the other models. These findings demonstrate the usability of pupillometric measurements and machine learning to predict cognitive workload and benefit professionals in various domains with complex visuospatial tasks.
Event Type
Poster
TimeTuesday, October 14th5:30pm - 6:30pm CDT
LocationRiverside East
Similar Presentations