Scholarlink

<p>Providing expert feedback for surgical skill acquisition is crucial but resource-intensive. Artificial intelligence (AI) offers a scalable solution, yet its effectiveness requires rigorous comparison with traditional methods. This prospective randomized controlled trial evaluated the impact of an AI tutor versus expert human instruction and self-directed learning on the foundational laparoscopic skill development of 124 novice medical trainees. Over a 4-week standardized curriculum, participants were randomized to receive feedback from: (1) an AI tutor utilizing video analysis (MetaGP-SurgEd), (2) expert human instructors, or (3) self-directed learning (control). Primary outcomes were post-training, expert-blinded Objective Structured Assessment of Technical Skills (OSATS) scores and simulator-derived metrics. Secondary outcomes included self-reported emotions and cognitive load. Results showed that both the AI Tutor and Human Tutor groups demonstrated significantly superior performance compared to the control group across all primary outcome measures (<i>p</i> &lt; 0.001). Crucially, no statistically significant differences were found between the AI Tutor and Human Tutor groups on any performance metric (<i>p</i> &gt; 0.05). Both feedback groups also reported more positive emotional states and more efficient cognitive processing, with lower extraneous and higher germane cognitive loads than controls. In conclusion, for novice medical trainees, AI-driven tutoring was as effective as expert human instruction for acquiring foundational laparoscopic skills and superior to self-learning. These findings validate AI tutors as a viable and scalable tool to complement traditional expert feedback in surgical education.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evaluating the impact of AI-tutoring versus expert human instruction on surgical skills in medical students

  • 期刊论文

  • 发表日期：

    2025年10月10日

查看全文
  • Yili Sun,
  • Fei Liu

摘要

Providing expert feedback for surgical skill acquisition is crucial but resource-intensive. Artificial intelligence (AI) offers a scalable solution, yet its effectiveness requires rigorous comparison with traditional methods. This prospective randomized controlled trial evaluated the impact of an AI tutor versus expert human instruction and self-directed learning on the foundational laparoscopic skill development of 124 novice medical trainees. Over a 4-week standardized curriculum, participants were randomized to receive feedback from: (1) an AI tutor utilizing video analysis (MetaGP-SurgEd), (2) expert human instructors, or (3) self-directed learning (control). Primary outcomes were post-training, expert-blinded Objective Structured Assessment of Technical Skills (OSATS) scores and simulator-derived metrics. Secondary outcomes included self-reported emotions and cognitive load. Results showed that both the AI Tutor and Human Tutor groups demonstrated significantly superior performance compared to the control group across all primary outcome measures (p < 0.001). Crucially, no statistically significant differences were found between the AI Tutor and Human Tutor groups on any performance metric (p > 0.05). Both feedback groups also reported more positive emotional states and more efficient cognitive processing, with lower extraneous and higher germane cognitive loads than controls. In conclusion, for novice medical trainees, AI-driven tutoring was as effective as expert human instruction for acquiring foundational laparoscopic skills and superior to self-learning. These findings validate AI tutors as a viable and scalable tool to complement traditional expert feedback in surgical education.