...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Quality Control and Applied Statistics >Optimal, two-stage, adaptive enrichment designs for randomized trials, using sparse linear programming
【24h】

Optimal, two-stage, adaptive enrichment designs for randomized trials, using sparse linear programming

机译：Optimal, two-stage, adaptive enrichment designs for randomized trials, using sparse linear programming

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相关主题

摘要

Consider a randomized trial of a new treatment versus control when the population is partitioned into two subpopulations. Standard designs may have only low power if the benefit of the treatment is on one subpopulation. In such cases, adaptive enrichment designs are useful. In a trial, it was found that there was a greater likelihood of benefitting the subpopulation that enter the trial with a cutaneous lesion compared to those with a non-cutaneous lesion. A two-stage adaptive enrichment design was implemented. In stage 1, enrollment was from the combined population. At the end of stage 1, there were four possible choices: 1. Continue enrolling the combined population, second, 2. Continue enrolling the combined population but having a larger sample size, 3. Enroll only those in the subpopulation with cutaneous lesions, or 4. End the trial. The objective is to select one of the four options based on results of stage 1 with a with minimum number of participants and trial duration as short as possible. The article addresses problems where minimum power is required to detect treatment benefits in the combined population as well as the subpopulation. The two-stage enrichment design consists of decision rule to modify the enrollment at the end of stage 1 and multiple-testing procedure at the end of stage 2 with a goal of minimizing the expected sample size taking into account the power and type I error. This non-convex optimization problem is not computationally feasible directly. To overcome this, the original optimization problem is approximated by a sparse linear program while simultaneously optimizing the decision rule and multiple-testing procedure in stage 2. The proposed designs control the family-wise type I error rate, which is mandatory in many countries. The subpopulation has to be defined before the trial starts and is based on prior trial data and disease-specific knowledge. The optimized designs improve power compared with standard designs. With more than two stages, the method becomes computationally infeasible.

著录项

获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号