自适应临床试验设计的近期进展及展望  

Response-Adaptive Randomization in Clinical Trials:Recent Advances and Future Perspectives

作　　者：高敬雅 马维 张绍洪 胡飞芳[4] GAO Jing-ya;MA Wei;CHEUNG Siu-hung;HU Fei-fang(School of Statistics,Renmin University of China,Beijing 100872,China;Institute of Statistics and Big Data,Renmin University of China,Beijing 100872,China;Department of Statistics and Data Science,Southern University of Science and Technology,Shenzhen 518055,China;Department of Statistics,The George Washington University,Washington DC,20052,US)

机构地区：[1]中国人民大学统计学院,北京100872 [2]中国人民大学统计与大数据研究院,北京100872 [3]南方科技大学统计与数据科学系,广东深圳518055 [4]乔治华盛顿大学统计系,美国华盛顿20052

出　　处：《数理统计与管理》2020年第4期595-610,共16页Journal of Applied Statistics and Management

基　　金：国家自然科学基金(11731011,11801559)的支持。

摘　　要：由于具有在伦理道德、有效性和灵活性等方面的优势,自适应设计在临床试验中受到越来越广泛的关注。本文首先介绍了两类主要的自适应设计,随机化罐子模型和基于序贯估计的方法。随后,本文着重介绍了自适应设计的近期进展:(1)自适应设计中的序贯监测问题;(2)自适应设计在响应变量为生存数据、复发事件数据和有序数据的试验中,以及在非劣效试验中的应用。最后我们探讨了自适应设计中尚需进一步解决的问题以及今后可能的研究方向。Because of its three advantages:Ethical,efficient and flexible,response-adaptive randomization(RAR)has recently attracted a great deal of attention in clinical research.In this paper,we first introduce two main types of RAR,namely,randomized urn models and the sequential-estimation model.Then we focus on the recent advances of RAR in the literature:(i)sequential monitoring of clinical trials based on RAR;and(ii)the applications of RAR for survival data,recurrent event data,and ordinal data and in non-inferiority trials.Lastly,we discuss a few unsolved issues and suggest some future research directions.

关 键 词：自适应设计 随机化罐子模型 双重适应性有偏硬币设计 最优分配 

分 类 号：O212[理学—概率论与数理统计]