应用统计硕士论坛

题      目： Stochastic Modeling and Optimal Adaptive Design of Clinical Trials

报  告  人：易雁青  教授   (邀请人：刘秀湘 )

                                   Memorial University of Newfoundland， Canada

时      间：2024-03-07 14：30--15：30

地      点：学院东楼507

报告人简介:

         Yanqing YI, Ph.D. is a professor of biostatistics in the Medicine department at Memorial University of Newfoundland, Canada. His major research interests include  statistical design and analysis for clinical trials, dependent data analysis, survival data analysis,stochastic modeling and Markov processes, statistical modeling and computation for complex data.

  摘要:

       Response adaptive designs use the information collected during a clinical trial to modify the randomization probabilities in order to allocate more patients to the potential better treatment. The designs have ethical advantages over the traditional half-half randomization designs, but it introduces a dependent structure in data, which may result in a statistical power loss. In this talk, we will discuss the trade-off between the ethical gain and the loss of statistical power to explore the optimal design of response adaptive designs of clinical trials. We formulate the
randomized treatment allocation process in an adaptive clinical trial as a stochastic sequential decision problem. When the information of previous treatment allocations and associated responses are summarized with sufficient statistics for unknown parameters, the decision process of treatment randomization becomes a Markov process on which a span-contractor operator is defined. It is proven that the average reward under the policy identified from the span-contractor operator converges almost surely to the optimal value. An algorithm is proposed to approximate the optimal value under the average reward criterion. Numerical
results reveal that the sequential procedure based on the controlled Markov process shows superior ethical advantage and at the same time produces good statistical power for large sample sizes such as 200 or larger.

        欢迎老师同学们积极参与交流！