时间：2024年3月13日 10:00-11:00
地点：普陀校区理科大楼A302
报告人：Richard Chappell University of Wisconsin -Madison
主持人：华东师范大学 徐进教授
摘要：
Textbooks describing how to analyze time-to-event outcomes in clinical trials tend to list a limited range of topics. Differences are often quantified using hazard ratios from the Cox model and its associated score, the log-rank test. Weighted rank tests may be presented, along with comparisons of landmarks and quantiles. All these have their disadvantages in terms of interpretation, convenience, and/or power. Furthermore, cancer immunotherapy and many other treatments in a variety of diseases can have delayed effects causing pairs of curves to diverge after months or years of followup. In such cases the log-rank test's power will be low and the associated hazard ratio estimate uninterpretable. Rank tests with increasing weights suffer from the paradoxical property of rewarding early failures. I will discuss various available alternatives including some which are quite new.
报告人简介：
Rick Chappell, Ph.D., is Professor in the Departments of Biostatistics and Medical Informatics and of Statistics at the University of Wisconsin Madison. His methodologic research is in the areas of Phase I clinical trials, non-inferiority trials, designs for randomized clinical trials, models for radiation biology, and survival analysis. He is Fellow of the American Statistical Association and the Society for Clinical Trials, has served as President of the Society for Clinical Trials, and is serving or has served on several FDA advisory committees and a large number of data safety monitoring boards for the NIH, VA, and industry.