AIを活用した看護学生の共感力向上VRシミュレーション：ランダム化比較試験における自動採点と評価者間信頼性

Empathy is a fundamental nursing competence that supports effective communication and high-quality care. In obstetric settings, the emotional intensity and unpredictability of childbirth place heightened demands on advanced empathic communication. Yet conventional teaching methods often lack immersion and interactivity, limiting students' capacity to respond empathically in high-pressure clinical situations.

To evaluate the effectiveness of the Virtual Reality Integrated Artificial Intelligence Empathic Communication Simulation (VR-AI-ECS) in enhancing nursing students' empathic communication, and to examine the concordance between AI-generated and educator ratings.

A randomised controlled mixed-methods design assigned students to an experimental group (VR-AI-ECS, n = 40) or a control group (360° video with standardised patient interaction, n = 40). The experimental group experienced first-person simulations incorporating AI-guided dialogue and real-time feedback. Empathic communication was measured at T0, T1, and T2 using the Empathic Communication Performance Rating Scale, evaluated by both AI and educators. Quantitative data were analysed using generalised estimating equations and intraclass correlation coefficients (ICC), and qualitative insights were gathered from focus groups.

The experimental group demonstrated significantly higher empathic communication scores at T1 (p < .001, Cohen's d = 0.96) and T2 (p = .010, Cohen's d = 0.60). AI and educator ratings were strongly correlated (ICC = 0.81). Students reported that dual perspectives and AI feedback enhanced their awareness and communication.

The VR-AI-ECS effectively enhances empathic communication and self-reflection among nursing students. AI-generated scoring shows strong alignment with expert judgement, supporting its application in scalable, standardised, and personalised empathy training.