LAUSR

Teachers’ engagement in online learning is a key factor in improving the effectiveness of online teacher training. This paper introduces a multimodal learning analytics approach that uses data on brain waves, eye movements and facial expressions to predict in-service teachers’ engagement and learning outcomes in online synchronous training. This study analyzed to what extent the unimodal and multimodal data obtained from the in-service teachers (n = 53) predict their learning outcomes and engagement. The results show that models using facial expressions and eye movements data had the best predictive performance on learning outcomes. The performance varied on teachers’ engagement: the multimodal model (integrating eye movements, facial expressions, and brain wave data) was best at predicting cognitive engagement and emotional engagement, while the one (integrating eye movements and facial expressions data) performed best at predicting behavioral engagement. At last, we applied the models to the four stages of online synchronous training and discussed changes in the level of teacher engagement. The work helps understand the value of multimodal data for predicting teachers’ online learning process and promoting online teacher professional development.