应数学系概率统计与运筹控制研究所刘伟邀请，美国食品药品管理局数理统计师张志伟将于7月1日-7月8日来公司进行讲学活动，欢迎感兴趣的师生参加！

报告题目：

（1）Treatment effect calibration, with application to non-inferiority trials I

（2）Treatment effect calibration, with application to non-inferiority trials II

（3）A potential outcomes framework for personalized medicine I: assessing treatment effect heterogeneity

（4）A potential outcomes framework for personalized medicine II: evaluating predictive biomarkers

报告时间： 7月5日上午9:00-10:30 ；10:40-12:10 

                          7月7日上午9:00-10:30 ；10:40-12:10 

报告地点：格物楼503

报告摘要：

In comparative effectiveness research, it is often of interest to calibrate treatment effect estimates from a clinical trial to a target population that differs from the study population. One important application is an indirect comparison of a new treatment with a placebo control on the basis of two separate clinical trials: a non-inferiority trial comparing the new treatment with an active control and a historical trial comparing the active control with placebo. Although ad hoc discounting methods are frequently used in practice, recent research on non-inferiority trials has focused on systematic calibration methods, including an outcome regression method based on a regression model for the outcome and a weighting method based on a propensity score model. Most recently, we developed methods based on a conditional effect model as well as doubly robust, locally efficient methods. This presentation will give an introduction to non-inferiority trials, describe the various calibration methods, compare them in a simulation study, and illustrate them with an HIV example.

Personalized medicine is basically an attempt to explore the heterogeneity of treatment effects (HTE) and utilize predictive biomarkers for treatment selection. The standard approach to assessing HTE and evaluating predictive biomarkers is based on a subgroup or regression analysis with focus on the treatment-by-marker interaction. Such an analysis is informative about the heterogeneity between subpopulations but not within. It is arguably more informative to assess HTE in terms of individual treatment effects, which can be defined using potential outcomes. Furthermore, under a potential outcome framework, a predictive biomarker can be formally defined as a predictor for a desirable treatment benefit and evaluated using familiar concepts in prediction and classification. While intuitively appealing, the potential outcomes framework is challenged by the lack of complete identifiability (because each patient can receive only one treatment in a typical study). One way to deal with the identification problem is to assume monotonicity of potential outcomes, with one treatment dominating the other in all patients. Motivated by HIV trials that do not generally follow the monotonicity assumption, we propose a different identification approach based on covariates and random effects. Under the proposed approach, the parameters of interest can be identified by assuming conditional independence of potential outcomes given observed covariates, and a sensitivity analysis can be performed by incorporating an unobserved random effect that accounts for any residual dependence. The proposed methods are illustrated with real data from HIV trials.

报告人简介：张志伟博士为美国食品药品监督管理局优秀的数理统计师，其具有在美国高校、美国国立健康研究院等著名科研院所的工作经历。张博士在统计理论研究及实践应用方面均有较深的造诣，其研究范围涵盖了缺失数据、因果推断、观测性研究、纵向数据等多个领域，目前已发表或接受的SCI文章超过60篇，其中多篇刊登于诸如Journal of the American Statistical Association, Biometrics等顶级统计学杂志上.