Model-Based Designs Considering Toxicity Alone
Model-Based Designs Considering Toxicity Alone
复制标题
DOI：
10.1007/978-4-431-55585-8_3
复制
发表时间：
2019-01-01
期刊：
影响因子：
--
通讯作者：
Matsui, Shigeyuki
中科院分区：
文献类型：
--
作者：
Daimon, Takashi;Hirakawa, Akihiro;Matsui, Shigeyuki
肿瘤学的I期试验设计旨在确定感兴趣的药物的最大耐受剂量。传统上，I期试验的剂量探索设计可分为基于规则/算法或模型的设计。在Chap. 2，我们关注基于规则的设计;在本章中，我们关注基于模型的设计。基于模型的设计假设单调剂量-毒性关系的某种统计模型，以在剂量间借用强度，并源于连续再评估方法（CRM）。本章详细介绍了CRM的概念、理论、特性和优缺点，并概述了与CRM相关或构成该方法扩展版本的设计。
Phase I trial designs in oncology aim to determine the maximum tolerated doses for agents of interest. Traditionally, dose-finding designs for phase I trials can be classified as rule-/algorithm- or model-based designs. In Chap. 2, we focused on rule-based designs; in the present chapter, we focus on model-based designs. Model-based designs assume a certain statistical model for the monotonic dose-toxicity relationship to borrow strength across doses, and originate from the continual reassessment method (CRM). The present chapter provides a detailed description of the concepts, theories, properties, and advantages and disadvantages of the CRM, and also overview designs that are related to the CRM or that constitute extended versions of this method.