Using learning analytics to identify engagement trajectories associated with clinical performance gains in virtual simulation

Background Although virtual simulation is widely used in nursing education, how students engage with repeated practice remains poorly understood. Traditional evaluations based on pre- and post-test scores may overlook important behavioural learning patterns. This study aims to identify distinct engagement trajectories in virtual simulation and examine their association with performance improvement. Methods A retrospective observational study analysed log data from 330 nursing students completing a severe limb trauma simulation. Trajectory analysis included 148 students with two or more attempts. Behavioural indicators included attempts, error reduction, completion time change and total learning time. K -means clustering identified engagement trajectories and analysis of variance (ANOVA) compared clinical performance gains across groups. Results Three engagement trajectories emerged: High Engagement (n = 39), Moderate Engagement (n = 49) and Low Engagement (n = 60). Clinical performance gains differed significantly across trajectories (F = 414.83, p < .001), with the High-Engagement group showing the greatest improvement. Students who repeated practice initially scored lower but ultimately achieved higher simulation performance scores than single-attempt peers. Conclusions Engagement trajectories within virtual simulation are strongly associated with learning outcomes. Learning analytics may help identify students needing targeted instructional support.