报告名称：Kernel quantileregression for semiparametricpartially lineartime-varying-coefficient modelbased on ahistoryprocess oflongitudinal data
报告人：刘秀芳 太原理工大学
报告时间：2026年3月11日 上午10:30--11:30
会议地点：伍卓群楼第一报告厅
校内联系人：李聪li_cong@jlu.edu.cn
报告摘要：This study delves into kernel quantile regression estimation fora semiparametric partiallylinear time-varying-coefficient model, which incorporates a history process withtime-dependentcovariatesandaright-censoredtime-to-eventvariable.Weproposeathree-stageapproachtoconstructtheestimatorsoftheparametricportionandnonparametrictime-varying-coefficientfunction for this model, in viewof inverseprobabilityof censoringweighting(IPCW)technique.Additionally, we offer a procedure for variable selection among the time-dependent covariates inthe parametric segment through the use of an adaptive LASSO penalty. The paperestablishestheasymptotic normality of the proposed estimators and demonstrates thatthepenalizedestimatorspossess the oracle property. A numerical simulation isimplementedtoevaluatetheperformanceoftheproposedestimators.Eventually,weapplythedevelopedmethodtoanalyzemedicalcostdatafromamulticenterautomaticdefibrillatorimplantationtrial(MADIT)toillustrateitspracticalutility.
个人简介：
刘秀芳，太原理工大学教师，博士毕业于吉林大学，2018年在加拿大里贾那大学访学一年。刘老师研究兴趣包括时间序列分析、生物统计等， 以第一作者或通讯作者在包括Journal of Multivariate Analysis, Statistical Methods in Medical Research等国际知名统计期刊发表SCI论文12篇。
