大语言模型在药品监管中的应用实践与思考  

Practices and Reflections on the Application of Large Language Models in Drug Regulation

作　　者：陈锋[1] 吴欣然 CHEN Feng;WU Xin-ran(Center for Information,NMPA)

机构地区：[1]国家药品监督管理局信息中心

出　　处：《中国食品药品监管》2025年第3期4-13,共10页China Food & Drug Administration Magazine

摘　　要：近年来人工智能(AI),尤其是大语言模型技术的快速发展,为药品监管领域带来了新的机遇与挑战。本文结合国家药监局信息中心的初步实践,探讨了药品监管领域应用大语言模型的基础考量,提出了一体化建设框架与实践路径。文章首先分析了数据、算力和算法三大基础要素的重要性,梳理了数据质量、算力成本和算法适用性在模型应用中的关键作用;提出了一体化建设框架,建议通过国家与省级药品监管部门的协同合作,构建统一且可扩展的大语言模型应用体系,避免重复建设和资源浪费;最后展示了大语言模型在药品注册形式审查场景中的具体案例及实际应用成效,并展望了未来药品监管大语言模型在智能化水平、安全性、合作平台和个性化服务开发等方面的发展前景。In recent years,the rapid development of artificial intelligence(AI)technology,particularly large language models(LLMs),has brought new opportunities and challenges to the field of drug regulation.This paper,based on the preliminary practices of the Center for Information,NMPA,explores the fundamental considerations for applying LLMs in drug regulation and proposes an integrated construction framework and practical pathways.The article first analyzes the importance of three foundational elements:data,computing power,and algorithms,emphasizing the critical roles of data quality,computing cost,and algorithm suitability in model applications.Next,it proposes an integrated construction framework,suggesting collaborative efforts between national and provincial drug regulatory authorities to build a unified and scalable LLM application system that avoids redundant construction and resource waste.Finally,through a case study on the intelligent review of drug registration documents,the paper demonstrates the practical outcomes of LLM applications and envisions future developments in intelligent systems,safety measures,collaborative platforms,and personalized service development in drug regulation.

关 键 词：药品监管 人工智能 大语言模型 一体化建设框架 实践案例 

分 类 号：R95[医药卫生—药学]