摘要
目的:为病理图像人工智能分析软件提供临床试验思路。方法:分析已发布的关于软件类产品临床试验设计的指导原则及已经批准的人工智能辅助诊断类产品的试验情况,结合病理图片人工智能分析软件特点,对该类产品临床试验进行深入研究。结果:在临床试验合规性考虑、试验设计、入组人群、样本量、统计分析、特殊考虑等方面形成病理图像人工智能分析软件的临床试验思路。结论:病理图像人工智能分析软件的临床评价,有助于该类产品更加科学合理地完成临床验证,促进产品尽快上市。
Objective To provide clinical trial ideas of AI analysis software for pathological images.Methods Through analyzing the published guidelines on clinical trial design of software products and the approved clinical trials of AI-assisted diagnosis products,we made an in-depth study of the clinical trial of these products based on the features of AI analysis software for pathological images.Results The clinical trial idea of AI analysis software for pathological images was formed in the aspects of clinical trial compliance consideration,trial design,enrolled population,sample size,statistical analysis,special consideration and so on.Conclusion Clinical evaluation of AI analysis software for pathological images is helpful to conduct clinical validation more scientifically and reasonably,and promotes the product to be marketed as soon as possible.
作者
徐超
郑生伟
方丽
李冉
吕允凤
XU Chao;ZHENG Shengwei;FANG Li;LI Ran;LYU Yunfeng(Center for Medical Device Evaluation,National Medical Products Administration,Beijing 100081,China)
出处
《中国数字医学》
2023年第5期23-26,31,共5页
China Digital Medicine
基金
科技创新2030-“新一代人工智能”重大项目(2020AAA0105000,2020AAA0105001)。
关键词
病理图像
人工智能
临床评价
Pathological images
Artificial intelligence(AI)
Clinical evaluation