Zamora Tomas, Salas Paulina, Zuñiga Sebastian, Botello Eduardo, Andia Marcelo E
Department of Orthopaedic Surgery. Pontificia Universidad Catolica de Chile, Santiago, Chile.
i-HEALTH Millennium Institute for Intelligent Healthcare Engineering, Santiago, Chile.
J Clin Orthop Trauma. 2025 Aug 7;69:103161. doi: 10.1016/j.jcot.2025.103161. eCollection 2025 Oct.
Generative artificial intelligence (AI), particularly large language models (LLMs), has emerged as a transformative technology across all medical specialties, including musculoskeletal (MSK) oncology. These models, such as ChatGPT and others, can process natural language, synthesize vast amounts of information, and generate contextually relevant outputs that resemble human communication. In orthopedic oncology, LLMs show promise in facilitating literature reviews, enhancing patient education, and supporting clinical decision-making by analyzing multidimensional data while providing improved logic-based reasoning. Additionally, they can assist in radiological and pathological workflows by interpreting imaging reports and drafting diagnostic summaries, thereby increasing efficiency and accuracy. In the near future, they are expected to aid in real-time patient follow-up and counseling, information transfer, efficient diagnostics, and even continuous surgical education and assistance. Despite their potential, challenges such as the risk of inaccuracies and biases, as well as the necessity for continuous supervision, warrant a cautious and responsible integration into clinical practice. This narrative review examines the current applications of LLMs in MSK oncology, their limitations, and their future potential in shaping precision medicine and equitable healthcare delivery.
生成式人工智能（AI），尤其是大语言模型（LLMs），已成为包括肌肉骨骼（MSK）肿瘤学在内的所有医学专科的变革性技术。这些模型，如ChatGPT等，能够处理自然语言，整合大量信息，并生成类似于人类交流的上下文相关输出。在骨科肿瘤学中，大语言模型在促进文献综述、加强患者教育以及通过分析多维数据支持临床决策并提供改进的基于逻辑的推理方面展现出前景。此外，它们可以通过解读影像报告和起草诊断总结来协助放射学和病理学工作流程，从而提高效率和准确性。在不久的将来，预计它们将有助于实时患者随访与咨询、信息传递、高效诊断，甚至持续的外科手术教育与协助。尽管具有潜力，但诸如不准确和有偏差的风险等挑战，以及持续监督的必要性，都需要谨慎且负责任地将其整合到临床实践中。本叙述性综述探讨了大语言模型在MSK肿瘤学中的当前应用、局限性以及它们在塑造精准医学和公平医疗服务方面的未来潜力。
