期刊
BIOMETRICS
卷 58, 期 3, 页码 671-674出版社
BLACKWELL PUBLISHING LTD
DOI: 10.1111/j.0006-341X.2002.00671.x
关键词
continual reassessment method; indifference interval; model sensitivity
The continual reassessment method (CRM) is a sequential design used in phase I cancer trials to determine the maximal dose with acceptable toxicity. It has been established that the CRM is consistent under model misspecification but not generally. When the method does not converge to the target percentile, some dose-response models will be more sensitive than others in terms of how close the converged recommendation is to the target. In this article, we interpret the main condition under which the CRM is consistent and apply it to evaluate the sensitivity of the model used with the CRM. The technique presented is found to be a useful supplement to simulation when planning a phase I trial.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据