Effect of a generative AI-supported ERAS case-simulation package on nursing students’ knowledge and self-reported 21st-century skills: a randomized controlled trial

To examine the effects of an AI-supported case simulation (ERAS protocol) on nursing students’ gynecologic oncology care knowledge and 21st-century skills. Evidence on structured generative AI-supported case simulations in nursing education remains limited. A single-center, parallel-group randomized controlled trial. Seventy-six third-year undergraduate nursing students at a public university in western Türkiye were randomized to the AI intervention ( n = 38) or control group ( n = 38), and 70 participants (35 per group) were included in the final analyses. The intervention group completed a multicomponent AI-supported ERAS case-simulation package, whereas the control group received a 35–40 min online case-based session on the same ERAS topic delivered by the study authors using a PowerPoint presentation and question-and-answer discussion. The two primary outcomes were ERAS nursing-care knowledge, assessed using the ERAS Protocol Nursing Care Knowledge Test, and total self-reported 21st-century skills; the individual skill dimensions were examined as exploratory secondary outcomes. Outcomes were assessed immediately before and after the educational activities and analyzed using a 2 × 2 mixed-design ANOVA. Significant Time × Group interactions were observed for ERAS nursing-care knowledge ( F (1, 68) = 92.908, p < .001, partial η² = 0.577) and total self-reported 21st-century skills ( F (1, 68) = 33.005, p < .001, partial η² = 0.327). The intervention group showed a large immediate between-group difference on the study-specific ERAS knowledge test, together with greater changes in several self-reported skill dimensions. However, self-reported critical thinking and problem-solving scores decreased, with a greater decline in the intervention group. The multicomponent AI-supported ERAS case-simulation package was associated with a large immediate between-group difference on the study-specific ERAS knowledge test and changes in several self-reported 21st-century skill dimensions. Because the intervention combined generative AI, interactive case-based learning, immediate feedback, structured scenario progression, and AI-generated visual materials, the findings reflect the educational package as a whole rather than the independent effect of generative AI. Longer-term effects and transfer to objective clinical performance remain to be established. ClinicalTrials.gov, NCT07183202. Registered on August 31, 2025. Retrospectively registered.