摘要
随机对照试验在临床试验中应用广泛,针对随机对照试验的样本分配是实际应用中需要考虑的重要问题.然而对于因果推断框架下的随机对照试验,传统样本分配方法没有考虑潜在结果相关性.为了解决这个问题,文章提出了一种考虑敏感性参数的样本分配方法,并结合自适应设计给出了常用因果效应的估计.模拟研究验证了该方法在潜在结果独立性假设不成立时也可以获得较其他常用分配方法更准确的因果效应估计.
The randomized controlled trial,(RCT)is widely used in clinical trials,and sample allocation for the RCT is an important issue in practical applications.However,for the RCT under the causal inference framework,traditional sample allocation methods did not consider the correlation of potential outcomes.To solve this problem,we propose a sample allocation method considering the sensitivity parameter,and combine it with a response-adaptive randomization design to estimate causal effects.Simulation studies have verified that our method can obtain more accurate estimators of causal effects than other commonly used allocation methods when the independence assumption does not hold.
作者
王宇航
WANG Yuhang(School of Mathematics and Statistics,Beijing Institute of Technology,Beijing 100081)
出处
《系统科学与数学》
CSCD
北大核心
2023年第12期3324-3338,共15页
Journal of Systems Science and Mathematical Sciences
关键词
因果效应
敏感性分析
样本分配
自适应设计
Causal effect
sensitivity analysis
sample allocation
response-adaptive randomization design
