报告承办单位: 数学与统计学院

报告内容: A Multiply Robust Estimation in Learning Optimal Individualized Treatment Regimes

报告人姓名: 朱文圣

报告人所在单位: 东北师范大学

报告人职称/职务及学术头衔: 教授，博士生导师

报告时间:2026年10月25日10:00-12:00

报告地点:长沙理工大学云塘校区理科楼A-212

邀请人：朱恩文

报告摘要：A core objective of personalized medicine is to identify optimal individualized treatment regimens whose recommendations rely on patients’ unique characteristics to maximize expected clinical outcomes. Traditionally, doubly robust methods provide double protection for consistent estimation against model misspecification. However, such methods permit only one propensity score model and one outcome model, and the requirement that one of the two models is correctly specified is restrictive in practical applications. To further improve the robustness of estimates beyond conventional doubly robust estimators, we develop a multiply robust value function estimator by constructing multiple candidate models for both the propensity score and the outcome. We prove that this estimator remains consistent if at least one model within either the propensity score candidate set or the outcome candidate set is correctly specified. Extensive simulation studies demonstrate that our method outperforms competing approaches. We further illustrate the utility of the proposed method via an analysis of data from the China Rural Hypertension Control Project.

报告人简介： 朱文圣，东北师范大学数学与统计学院教授、博士生导师。2006年博士毕业于东北师范大学，2008-2010年在耶鲁大学做博士后研究，2015-2017年访问北卡罗莱纳大学教堂山分校。研究方向为生物统计学及统计机器学习，研究成果发表于JASA, Biometrika, Biometrics、IEEE TGRS、《柳叶刀》等期刊，主持多项国家级科研项目。兼任中国现场统计研究会贝叶斯统计分会理事长、中国统计教育学会副会长，全国工业统计学教学研究会副会长。国家级一流本科课程负责人，吉林省拔尖创新人才。