首页|期刊导航|中国食品药品监管|优化人工智能医疗器械全生命周期监管:国际经验与中国路径

优化人工智能医疗器械全生命周期监管:国际经验与中国路径OA

Optimizing the Full Lifecycle Regulation of Artificial Intelligence Medical Devices:International Experience and China's Approach

中文摘要英文摘要

目的:探讨人工智能医疗器械监管的国际经验,分析我国监管体系现状与不足,提出可操作的优化路径,推动形成科学、灵活、透明的全生命周期监管体系.方法:采用比较研究方法,选取美国、欧盟、日本作为案例,从审批路径、前瞻性算法变更、风险分级和上市后监管4个维度综述人工智能医疗器械监管模式,并结合我国近年来医疗器械相关政策与实践进行系统梳理.结果:美国以灵活审评与创新支持为特征、欧盟强调风险预防与透明公开、日本在稳健审慎基础上引入前瞻性变更与数据驱动的上市后监测,3种模式在制度设计和实施效果方面各具优势.我国已形成由指导原则、标准和审评要点等支撑的人工智能医疗器械监管框架,并在真实世界数据及真实世界证据应用方面取得进展,但在算法更新的前瞻性变更路径、风险分级与技术要点的细化以及跨部门协同等方面仍有待完善.结论:我国应借鉴国际经验,完善风险分级制度、建立前瞻性算法变更的预审与动态变更机制、推动真实世界数据应用、健全跨部门协同机制、探索监管沙盒试点,且有必要探索构建以"审批-更新-使用-监测"闭环为核心的全生命周期监管体系,在保障安全有效的前提下提升迭代效率,形成可复制、可推广的监管实践.

Objective:This study aims to examine international regulatory experience in artificial-intelligence(AI)medical devices,assess the current status and limitations of China's regulatory framework,and propose actionable pathways toward a scientific,flexible,and transparent lifecycle regulatory system.Methods:A comparative research approach was adopted,selecting the United States,the European Union,and Japan as representative cases.Their regulatory models were reviewed across four dimensions:approval pathways,prospective algorithm change mechanisms,risk classification,and post-market surveillance,and analyzed in conjunction with China's recent medical device policies and practices.Results:The United States is characterized by flexible review mechanisms and strong support for innovation;the European Union emphasizes risk prevention and regulatory transparency;and Japan,while maintaining a prudent regulatory foundation,has introduced prospective change management and data-driven post-market monitoring.China has established an initial regulatory framework supported by guidance documents,standards,and technical review requirements,and has made progress in the application of real-world data/evidence(RWD/RWE).Nonetheless,gaps remain in developing prospective regulatory pathways for algorithm updates,refining risk classification and technical specifications,and strengthening interdepartmental coordination.Conclusion:China should draw on international experience to further refine its risk classification system,establish pre-review and dynamic management mechanisms for prospective algorithm changes,promote the systematic use of real-world data,strengthen cross-sector collaboration,and pilot regulatory sandbox initiatives.In particular,it is necessary to build a full lifecycle regulatory framework centered on a closed-loop model of"approval–update–use–monitoring",thereby improving iterative efficiency while ensuring safety and effectiveness and forming replicable and scalable regulatory practice.

李屹玲;赵琪;魏旗鹏;胡琳琳

中国医学科学院北京协和医学院卫生健康管理政策学院中国医学科学院北京协和医学院中国医学科学院北京协和医学院中国医学科学院北京协和医学院卫生健康管理政策学院

医药卫生

人工智能医疗器械全生命周期监管国际比较政策建议

artificial intelligencemedical devicesfull lifecycle regulationinternational comparisonpolicy recommendations

《中国食品药品监管》 2026 (3)

42-51,10

中国医学科学院医学与健康科技创新工程(2021-I2M-1-046)

10.3969/j.issn.1673-5390.2026.03.004

评论