临床试验中自适应设计的urn模型和RPW规则及其模拟.doc

文档介绍：临床试验中自适应设计的urn模型和RPW规则及其模拟
作者:王素珍, 夏结来,郑亮, 于莉莉,李婵娟,王陵
【摘要】目的: 探讨临床试验中设计驱动型自适应设计的优缺点及适用范围. 方法: 通过对基于概率罐子(urn)模型的随机化胜者优先(RPW)规则的原理分析得出期望分配公式,并据此进行蒙特卡罗(Monte Carlo)计算机模拟. 结果: 假定总体成功率分别为pA, pB的两处理临床试验,当pA, pB不相等且pA+pB<,采用RPW规则能够将更多的受试者分配给成功率较高的处理组. 结论: RPW规则可适用于临床试验中有一定响应延迟的情况,通过选择合适的RPW初始参数可减少样本方差,增加试验的稳定性.
【关键词】临床试验;自适应设计;模型,统计学;RPW规则;Monte Carlo模拟
【Abstract】 AIM: To probe into the advantages, disadvantages and application scope of designdriven adaptive design in clinical trials. METHODS: Expected allocation formula of randomized playthewinner(RPW) based on urn model in probability was deduced through principle analysis and Monte Carlo simulation was implemented according to it. RESULTS: Suppose the ess probabilities of the two treatment groups in clinical trials are pA and pB, pA is not equal to pB and pA+pB<, RPW rule can allocate more patients to a better treatment with higher ess probability. CONCLUSION: RPW rules can be applied to clinical trials with response delay. The variance can be reduced substantially and the stability of the clinical trial can be enhanced through appropriate selection of starting parameters of RPW.
【Keywords】 clinical trials; adaptive design; models, statistical; RPW rule; Monte Carlo simulation
传统的临床试验常采用平行对照设计,此方法操作简单,便于控制,统计效率较高. 然而,无论从药物的研发成本、速度和伦理道德的观点出发,该方法都不令人满意. 在临床试验设计中,人们希望尽量缩短药物的研发时间并将治疗效果好的处理尽可能多地分配给新药临床试验的志