Precision Education in Orthopedic Surgery: Impact of an AI-Driven RAG Model on Perioperative Nursing Students

Introduction: The complexity of orthopedic surgery causes the minds of perioperative nursing students to be stressed, and thus, they need to be better educated. The transformative potential of generative AI is real, and general-purpose Large Language Models (LLMs) are not domain-specific and hallucinate. The effect of an orthopedic surgical technology course on learning outcomes and student satisfaction will be analysed with the help of a domain-specific Retrieval-Augmented Generation (RAG) model, which is based on authoritative literature on surgery. Methods: 50 perioperative nursing students in two Tehran medical universities were selected to take part in a quasi-experimental non-equivalent pretest-posttest controlled group. The control group (n=25) was given conventional teaching, and the intervention group (n=25) was given RAG-based intelligent teaching. The Education Satisfaction Questionnaire and the Orthopedic Surgical Technology Knowledge Test were verified. Mann-Whitney U and Wilcoxon tests were done by SPSS. Results: The difference in learning was significant in the intervention group, and the effect size was greater (Mean Gain: 15.72 vs. 12.64, P < 0.001, r = 0.77). Team RAG-based system had a moderate-high user satisfaction (Mean: 72.48+7.56), which means that the technology was used successfully and positively influenced the user experiences. Discussion: The domain-specific RAG model might minimize cognitive load and grounded retrieval hallucinations as an intelligent tutoring support. These findings indicate that precise surgical teaching devices can enhance cognitive education and change education. This knowledge should be taken long-term through research and clinical use.