大语言模型重塑外科未来——大语言模型在外科领域的应用进展

Large Language Models Reshaping the Future of Surgery—Advances in the Application of Large Language Models in the Field of Surgery

摘要: 以ChatGPT、Gemini、DeepSeek、通义千问等为代表的大语言模型正蓬勃发展，其应用已渗透至医疗实践的各个领域，将深刻改变未来医院的格局。在胸外科、心脏病学、口腔外科、肾脏病学、骨科、胃肠病学和影像科学等领域，变革尤为迅速。大语言模型在辅助医学文档书写、提供临床决策支持、进行医学健康教育及患者围手术期管理等方面展现出巨大的应用潜力。本文综述了大语言模型在电子病例书写、临床辅助诊断、临床决策支持、患者健康管理、医学教育及科研论文撰写等多个外科相关场景的应用。大语言模型能够高效处理与分析大规模数据集，并具备出色的自然语言理解能力。然而，这些技术的应用仍存在局限性，如模型的“幻觉”现象、潜在的学术不端风险、临床过度依赖、误诊与治疗失误的可能性以及责任归属不清等问题。在充分利用大语言模型益处的同时，我们必须认识并解决这些伦理与实践挑战，以确保其在医学领域的应用是负责任且有效的。

Abstract: Large language models (LLMs), represented by ChatGPT, Gemini, DeepSeek, Tongyi Qianwen, and others, are flourishing. Their applications have penetrated various fields of medical practice and are poised to profoundly reshape the future landscape of hospitals. Transformations are occurring particularly rapidly in fields such as thoracic surgery, cardiology, oral surgery, nephrology, orthopedics, gastroenterology, and imaging sciences. LLMs demonstrate immense application potential in assisting with medical documentation, providing clinical decision support, conducting medical health education, and managing patients during the perioperative period, among other areas. This article reviews the applications of LLMs in various surgery-related scenarios, including electronic medical record writing, clinical auxiliary diagnosis, clinical decision support, patient health management, medical education, and scientific research paper writing. LLMs can efficiently process and analyze large-scale datasets and possess remarkable natural language understanding capabilities. However, the application of these technologies still has limitations, such as model “hallucination,” potential risks of academic misconduct, clinical over-reliance, possibilities of misdiagnosis and treatment errors, as well as unclear attribution of responsibility. While fully leveraging the benefits of LLMs, we must recognize and address these ethical and practical challenges to ensure their application in the medical field is responsible and effective.