Effects of ChatGPT-generated immediate feedback integrated into VR-based OSCEs on nursing students’ performance: a randomized crossover study

The Objective Structured Clinical Examination (OSCE) is a gold-standard assessment in nursing education, yet traditional and virtual reality (VR) formats often lack timely, individualized feedback, potentially limiting gains in higher-order competencies such as communication and decision-making. Advances in artificial intelligence (AI), particularly ChatGPT, offer opportunities to address this gap. To evaluate the impact of ChatGPT-generated immediate feedback integrated into VR-based OSCE stations on nursing students’ communication performance, clinical decision-making accuracy, and related learning outcomes. A randomized, two-period crossover trial was conducted among 65 final-year nursing students at a tertiary teaching hospital in China between March and May 2025. Participants completed two matched OSCE stations under VR-only and VR + ChatGPT conditions, with a one-week washout period between the two study phases. Primary outcomes included OSCE communication scores, clinical decision-making accuracy, and task completion time. Secondary outcomes comprised learning satisfaction, acceptance of AI-assisted feedback, academic and communication self-efficacy, reflection quality, and cognitive workload. Data were analyzed using paired statistical comparisons and linear mixed-effects models adjusting for sequence and period effects. Compared with VR-only, VR + ChatGPT significantly improved communication scores (mean difference MD = 3.38, 95% CI: 2.12–4.64, P < 0.001, Cohen’s d = 0.63) and decision accuracy (MD = 3.91%, 95% CI: 2.09–5.72, P < 0.001, Cohen’s d = 0.49), while reducing completion time (MD = − 0.78 min, 95% CI: − 1.12 to − 0.44, P < 0.001, Cohen’s d = − 0.62). VR + ChatGPT also yielded higher learning satisfaction (MD = 0.31, P < 0.001), greater reflection quality (MD = 0.62, P < 0.001), lower cognitive workload (MD = − 5.7, P = 0.001), and significant gains in academic and communication self-efficacy (both P < 0.001). No severe adverse events occurred. Effects were consistent across subgroups defined by prior VR experience and digital literacy. Integrating ChatGPT-generated immediate feedback into VR OSCE scenarios significantly enhances nursing students’ communication and decision-making performance, accelerates skill acquisition, and reduces cognitive strain, without compromising safety. Not applicable.