科研交流
您的位置>>首页 > 科研交流 > 学术交流 > 阅读正文
科研交流

数学学院、所系列学术报告(713场):美国哈佛大学Lee-Jen Wei教授

admin 发表于: 2017-06-29 08:33  点击:399
题 目:Utilizing Patients' Multiple Time-to-Event Outcomes to Estimate a Clinically Meaningful Treatment Effect in a Comparative Study-Looking for Totality of Evidence
报告人:美国哈佛大学Lee-Jen Wei教授
时 间:2017年7月5日9:00--11:00
地 点:数学楼629
摘要:In a clinical study to compare a new therapy with a control (standard care), suppose that for each patient, there are several time-to-event outcomes (say, time to heart attack, stroke, heart failure, death et al.). These outcomes jointly reflect the disease progression or burden over time. The conventional way to estimate the treatment effect (difference) is based on a composite endpoint, which is the time to the first event of those multiple outcomes. This approach does not take the disease burden over time into consideration. The question is how to construct an endpoint which consists of outcomes beyond the first event with a clinically meaningful interpretation. We will review various statistical procedures for handling this problem. We then discuss details about two model-free inference procedures illustrated with the data from cardiovascular clinical trials. These procedures avoid the competing risks issue and provide totality of evidence on the treatment effect.