基于两种处理方案随机化设计的一些渐近性质

发布时间：2018-08-05 20:01

【摘要】：本文探讨了药物临床实验中的两类随机化设计模型(坛子模型与有偏硬币模型)的统计特性,重点放在研究基于两种处理方案的随机化模型的数字特征与大样本性质,而且简单分析了平衡分配随机化设计模型与有偏硬币设计的关系。通过理论性质的探索,本文指出了平衡分配坛子模型,自适应随机化坛子模型,广义Friedman坛子模型,广义Polya坛子模型,有偏硬币设计等的收敛速度以及影响收敛速度的因素,发现Ehrenfest-Brillouin模型收敛的速度与参数α_1有关,我们可以通过选择参数α_1的值使其收敛速度比非对称的Erhrenfest坛子模型快。最后得出结论:Chen提出的Ehrenfest模型收敛到平衡的速度比其他的坛子模型的速度快。这些理论结果为具体临床试验中选取处理方案提供了一定理论依据。
[Abstract]:In this paper, the statistical characteristics of two kinds of randomized design models (jar model and biased coin model) in drug clinical trials are discussed. The emphasis is placed on the numerical characteristics and large sample properties of randomized models based on two treatment schemes. Furthermore, the relationship between the stochastic design model of balanced distribution and the design of biased coins is simply analyzed. Through the exploration of theoretical properties, this paper points out the convergence rate of balanced distribution jar model, adaptive random jar model, generalized Friedman jar model, generalized Polya jar model, offset coin design and so on. It is found that the convergence rate of the Ehrenfest-Brillouin model is related to the parameter 伪 _ Step1. By selecting the value of the parameter 伪 _ S _ 1, the convergence rate of the Ehrenfest-Brillouin model is faster than that of the asymmetric Erhrenfest jar model. Finally, it is concluded that the convergence to equilibrium of Ehrenfest model proposed by President Chen is faster than that of other jar models. These theoretical results provide a theoretical basis for the selection of treatment schemes in specific clinical trials.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2006
【分类号】：R311

【共引文献】