人工智能医疗监管与治理创新:突破现行规范框架的思考  

Innovation in artificial intelligence medical regulatory and governance:thoughts on breaking through the current normative framework

作　　者：陈炳澍 王玥[2] CHEN Bingshu;WANG Yue(Neusoft Medical Systems Co.,Ltd.,Shenyang 110179,China;School of Law,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]东软医疗系统股份有限公司,辽宁沈阳110179 [2]西安交通大学法学院,陕西西安710049

出　　处：《中国医学伦理学》2024年第9期1030-1036,共7页Chinese Medical Ethics

摘　　要：随着以深度学习和生成式人工智能为代表的人工智能应用的爆发式增长,人工智能技术已经进入医疗健康各个领域,包括临床诊疗、医院管理、科学研究、教育培训和健康管理等,为患者带来高效优质服务的同时,也涌现出一系列治理制度需要解决和探索的问题,尤其是在人工智能医疗产品监管方面。目前,各国监管机构大多以风险等级方式进行人工智能医疗产品监管,基本将人工智能产品定位为“辅助”性质,即需要医师签字确认自动判读结果并承担法律责任。世界卫生组织先后发布两项人工智能指南,旨在对卫生健康领域人工智能技术的开发、提供和部署提供指导框架,并在肺结核这类特殊疾病诊断上提出突破性“独立使用”的建议,美国监管机构也提出“软件预认证”的突破性思路以缩短审批流程并创新监管方式。尝试从技术治理制度创新的角度出发,借鉴国际相关经验,结合中国技术发展与实践情况,提出针对特定应用场景、风险程度较低人工智能产品逐步开放“独立”应用,以及对于尚未纳入监管的风险较高人工智能医疗产品进行“机构预认证”的治理思路,为制度创新提供一些可操作的路径。With the explosive growth of artificial intelligence(AI)applications represented by deep learning and artificial intelligence generated content(AIGC),AI technologies have deployed various fields of healthcare,including clinical diagnosis and treatment,hospital management,scientific research,education training,and health management.While bringing efficient and high-quality services to patients,it has also emerged a series of regulatory and governance issues that need to be resolved and explored,especially in the regulation of AI medical applications.Currently,regulatory agencies in various countries mostly regulate AI medical products based on risk levels,positioning AI products as“aided”in nature,that is,physicians are required to sign to confirm the results of the automated interpretation and take legal responsibility.The World Health Organization has released two successive AI guidelines,aiming to provide a guiding framework for the development,provision,and deployment of AI technology in the field of health,and to propose breakthrough“independent use”recommendations for the diagnosis of special diseases such as tuberculosis.The United States regulators have also proposed a creative idea of“software pre-certification”to shorten the approval process and innovate regulatory methods.Starting from the perspective of innovation in the technology governance systems,drawing on relevant international experiences,and combining them with China’s technological development and practice situation,this paper proposed a governance thinking of gradually opening up“independent”applications for AI products with specific application scenarios and lower risk levels,as well as conducting“institutional pre-certification”for high-risk AI medical products that have not yet included in regulation,so as to provide some operable paths for institutional innovation.

关 键 词：生成式人工智能 医疗器械 独立软件 计算机辅助检测 

分 类 号：R-052[医药卫生]