CLINICAL TRIAL METHODOLOGY IN PSYCHOPHARMACOLOGY

该项目继续开发临床试验设计和 精神药理学和其他学科的统计分析方法 精神病患者护理中使用的治疗方法。具体目标是： (1)制定剂量反应的设计和统计分析策略 和生物测定研究，以表征相对于 标准治疗。基于包括以下患者的模型 在任何剂量下无反应，谁在任何剂量下有反应或谁有反应取决于 关于他们的剂量是否足够，我们将开发方法来分析 “医生的选择”研究设计。递归分区方法 我们将研究如何确定具有共同相对效力的变量。 (2)开发临床试验设计和分析方法，导致 提供临床信息的措施。我们将开发一个推理框架 对于一般线性模型和双向模型的最优缩放技术 表，当数据来自有序类别时。我们将发展 有效设计与非参数和参数模型及估计量 具有临床信息量的参数。 (3)开发简单的设计和统计方法，以测试每个 联合疗法的成分是否有助于其疗效以及是否 这种组合是拮抗的、相加的或协同的。在 多变量组合问题，我们将研究新的假设公式 这进一步说明了每个组件都“贡献”的要求 对这种组合的影响。要求更高的替代品 将开发出可接受但要求较低的统一最佳方案。我们 我将研究找到一个简单的实验设计来测试的问题 两种或两种以上剂量时的剂量相加、协同或拮抗 组合、单个和多个端点以及两个或多个端点的组合 更多的毒品。 (4)开发稳健、最优的交叉和响应自适应临床 用于评估治疗和遗留效应的试验设计。我们会 为任意数量的治疗寻求最优交叉设计 有结转影响的时期。我们将制定应对措施 适应性设计，将治疗分配给患者，以增加功率 对于涉及最小统计量的假设测试情况，以及 最优交叉设计问题。

This project continues the development of clinical trial design and statistical analysis methodology for psychopharmacological and other treatments used in the care of the mentally ill. The specific aims are to: (1) develop design and statistical analysis strategies for dose-response and bioassay studies to characterize a test treatment relative to a standard treatment. Based on models that include patients who are unresponsive at any dose, who respond at any dose or who respond depending on whether their dose is adequate, we will develop methods to analyze "doctor's choice" study designs. Recursive partitioning approaches for identifying variables that share a common relative potency will be studied. (2) develop clinical trial designs and methods of analysis resulting in clinically informative measures. We will develop an inference framework for optimal scaling techniques for the general linear model and for two-way tables when the data arise from ordered categories. We will develop efficient designs and nonparametric and parametric models and estimators of parameters that are clinically informative. (3) develop simple designs and statistical methods for testing whether each component of a combination therapy contributes to its effect and whether the combination is antagonistic, additive or synergistic. In the multivariate combination problem, we will study new hypothesis formulations that further characterize the requirement that each component "contributes" to the combination's effects. Alternatives that are more demanding than admissible but less demanding than uniformly best will be developed. We will study the problem of finding a simple experimental design to test for dose additivity, synergy or antagonism when there are two or more doses of the combination, single and multiple endpoints and combinations with two or more drugs. (4) develop robust and optimal crossover and response adaptive clinical trial designs for estimation of treatment and carryover effects. We shall seek optimal crossover designs for an arbitrary number of treatments and periods when there are carryover effects. We will develop response adaptive designs to allocate treatments to patients so as to increase power for hypotheses testing situations involving a min statistic and in the optimal crossover design problem.