Cognitive Engagement Measurement in Online Learning Using Eye-Tracking Data and Back Propagation Neural Networks

• In Event: AERA Roundtable Session Sunday 1:45 pm - Gold 3
  In Roundtable Session: RT1: Measurement and Related Issues in Quantitative Research Designs, Analysis, and Simulation

Abstract

This study proposes an intelligent assessment model for online learners’ cognitive engagement using eye-tracking and a backpropagation neural network (BPNN). By collecting eye-tracking data from six areas of interest—including text, charts, and others—eighteen features were extracted as model inputs to predict low, medium, or high engagement levels.The BPNN, with higher overall accuracy (0.9382) and F1 score (0.9371), outperformed comparison models such as SVM, XGB, and LSTM across all metrics, demonstrating superior nonlinear modeling ability and prediction stability. This approach supports data-driven optimization of instructional strategies and provides a scalable solution for personalized cognitive monitoring in online education.

Authors

• jialin zhou, East China Normal University

• Zhan Chen, East China Normal University

• Yaofeng Xue, East China Normal University

• 扣琪 王, East China Normal University

• yiwei gou, East China Normal University