Concordance Matched Learning for Estimating Optimal Individualized Treatment Regimes

报告题目:Concordance Matched Learning for Estimating Optimal Individualized Treatment Regimes

主讲人:朱文圣

报告摘要: Precision medcine has drawn tremendous attention recently to account for signifcantheterogeneity in the response of diferent patients to the same treatment. The estimation of the optimal indivdualized treatment regime (TR) Is of great concem to precision medicine, which is aim to recommenda treatment regime based on patient spcific characteristics by maximizing the expected clinical outcome,! In recent staisical iteratures, there is a large and growing body of iferent statistical methods to estimateopimal Indviduallzed treatment reglmes. Most of the existing statistical methods are mainly focus on theestimation of optimal idividualized decision nules for the two categories of treatment options and relyheavily on data from randomized contolled trials. In this talk, we propose a machine leaming approach(CM-earning) to estimate optimal treatment regime from mutticategorical treatment options, which alowsfor more accurate assessment of individual treatment response and aleviation of confounding. More importanty. CM-loaning is doubly robust, fficlent and easy to interpret. Through a large number ol simulation studses, we demonstrate that CM-leaming outperforms existing methods. Lastly, the proposedmethod is ilustrated in an analysis of AIDS clinical trial data.

朱文圣简介:东北师范大学数学与统计学院教授、博士生导师,副院长.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年访问北卡罗来纳大学教堂山分校.中国现场统计研究会计算统计分会副理事长,数据科学与人工智能分会秘书长,中国概率统计学会副秘书长,吉林省现场统计研究会秘书长.研究方向为生物统计与精准医疗,在JASA、Test、Neurolmage、中国科学等杂志发表学术论文多篇,主持并完成国家自然科学基金项目,入选吉林省第七批数尖创新人才.

报告时间: 2020年11月17日下午4:30-5:30

报告地点:南湖校区教学科研楼609室