| 摘要: |
| 大语言模型(Large Language Models,LLMs)和多模态模型(Multimodal Models,MMLs)通过整合文本、图像、语音等多模态数据，为临床诊断、个性化治疗及慢性病管理提供了全新的技术支持。本文系统梳理了LLMs和MMLs的技术基础及其在临床医学中的应用场景，包括临床诊断与决策支持、个性化医疗、慢性病管理等领域，探讨了其在提升诊断准确性、优化治疗方案及改善患者健康管理等方面的潜力与局限性。同时，深入分析了LLMs和MMLs在医疗领域面临的技术挑战，包括模型泛化能力不足、可解释性与透明性欠缺、隐私与数据安全风险，以及与现有医疗系统的兼容性问题，并阐述了这些挑战对技术落地和推广的影响。最后，本文展望了模型优化、数据融合及隐私保护等方面的发展方向，提出通过技术创新与跨领域协作，推动人工智能(Artificial Intelligence,AI)技术在医学领域的深度应用，为提升医疗服务效率和质量提供参考。 |
| 关键词: 大语言模型 多模态模型 临床大数据 临床辅助决策 个性化医疗 |
| DOI：10.13656/j.cnki.gxkx.20250521.010 |
| 投稿时间：2025-01-10修订日期：2025-02-03 |
| 基金项目:广西高校中青年教师科研基础能力提升项目(2022KY0093)和广西壮族自治区卫生健康委西医课题(Z-A20230708，Z-A20240684)资助. |
|
| Application and Challenges of Large Language Models and Multimodal Models in Clinical Medicine |
|
ZOU Yuan, TAN Yuping
|
| (The Second Affiliated Hospital of Guangxi Medical University, Nanning, Guangxi, 530007, China) |
| Abstract: |
| Large Language Models (LLMs) and Multimodal Models (MMLs) integrate text,images,and audio data to provide innovative technical support for clinical diagnosis,personalized treatment,and chronic disease management.The technical foundations of LLMs and MMLs and their applications in clinical medicine,including clinical diagnosis and decision support,personalized treatment,and chronic disease management,are systematically reviewed.The potential and limitations of LLMs and MMLs in enhancing diagnostic accuracy,optimizing treatment plans,and improving patient health management are explored.Furthermore,the technical challenges faced by LLMs and MMLs in the medical domain,such as limited generalization capability,issues with interpretability and transparency,risks related to privacy protection and data security,and compatibility challenges with existing medical systems,are examined.These challenges are highlighted as key barriers to the implementation and widespread adoption of these technologies.Finally,the future directions in model optimization,data integration,privacy protection are prospected,and it is proposed that technological innovation and multidisciplinary collaboration are needed in advancing the application of Artificial Intelligence (AI) in medicine.The paper provides a reference for improving healthcare service efficiency and quality. |
| Key words: Large Language Models (LLMs) Multimodal Models (MMLs) clinical big data clinical decision support personalized medicine |
