数智时代医疗人工智能的潜在法律风险及规制措施  

Potential Legal Risks and Regulatory of Medical Artificial Intelligence in the Era of Digital Intelligence

作　　者：游俊哲 丁岱睿 You Junzhe;Ding Dairui(People's Procuratorate of Gusu District,Suzhou 215007,China;School of Law,Southeast University,Nanjing 211189,China)

机构地区：[1]江苏省苏州市姑苏区人民检察院,江苏苏州215007 [2]东南大学法学院,江苏南京211189

出　　处：《中国科技论坛》2025年第3期129-140,共12页Forum on Science and Technology in China

摘　　要：医疗人工智能的出现引发医师诊疗活动的数字化转型。规范层面政策性文件展现出的“鼓励创新与规制风险并行”的态度与现实层面医疗AI利弊相伴的特征,唤起医疗AI风险规制的内在需求。医疗AI在当前技术阶段造成的现实风险主要包括影像识别系统的判断偏差、医疗专家系统的计算失真以及医疗机器人的行为失控。试图从主体论出发规制风险的医疗AI主体资格肯定论以“往后必然产生独立意识”为由论证赋予其法律主体资格为时过早且不具有实际意义。应在医疗AI风险规制中引入比例原则,根据患者知情同意设置适当性原则的二元标准,以医疗疾病划分构建最小侵害原则的双重结构以及合比例原则的差异化审查机制,同时运用层次分析法量化分析AI类型、疾病严重程度、患者知情同意情况、AI运用损益情况对医疗人工智能应用限度的影响,以量化形式确定AI应用的可行性与风险性。The emergence of medical artificial intelligence has triggered the digital transformation of physician diagnosis and treatment activities.The policy documents at the normative level demonstrate an attitude of encouraging innovation while regulating risks,which,accompanied with the dual characteristics of advantages and disadvantages of medical AI in practice,arouses the internal demand of medical AI risk regulation.The practical risks caused by medical AI in the current technical stage mainly include the judgment deviation of image recognition system,the calculation distortion of medical expert system and the out of control of medical robot behavior.The affirmation theory of medical AI subject qualification,which attempts to regulate risks on the grounds that independent consciousness will inevitably emerge in the future,is too early and has no practical significance.The principle of proportionality should be introduced into the risk regulation of medical AI,and the binary standard of the principle of appropriateness should be set according to the informed consent of patients.The dual structure of the principle of minimum infringement and the differentiated review mechanism of the principle of proportionality should be constructed by the division of medical diseases.At the same time,the analytic hierarchy process was used to quantitatively analyze the impact of AI type,disease severity,informed consent of patients,and profit and loss of AI on the limit of medical artificial intelligence application,so as to determine the feasibility and risk of AI application in a quantitative form.

关 键 词：医疗人工智能 风险规制 比例原则 患者知情同意 层次分析法 

分 类 号：D912.1[政治法律—宪法学与行政法学]