题 目：Mixture Cure Model with both Semiparametric Components for Interval-Censored Data

报告人：刘晓玉（暨南大学）

Abstract：Mixture cure models (MCM) have become popular in clinical studies where a subset of subjects never experiences the failure event. Those models have been developed as an extension of the matching semiparametric survival models. However, most MCM’s developed in the literature often assumes the probability of susceptibility as a known transformation of the linear combination of covariates. In some practical cases, it is not realistic to assume that a covariate with continuous support has a linear effect on the susceptibility. Therefore, more complex models may be necessary. We propose a double semiparametric mixture cure model for interval censored data that allows one covariate to have nonlinear effect on the susceptibility. We provide a sieve semiparametric maximum likelihood estimator based on polynomial splines to estimate the parameters and the unknown functions in the model. We establish large sample properties of the proposed estimator. Through empirical evidence, we show that given large enough sample size, we do not lose much efficiency compared to the usual logit model for the probability of susceptibility and we gain the flexibility of a non-linear term. To illustrate the proposed method, we apply the proposed double semiparametric approach to data from a hypobaric decompression sickness study.

报告人简介：刘晓玉，现任暨南大学经济学院、统计与数据科学系助理教授。2015年毕业于东南大学数学学院，获统计学专业理学学士学位，2020年毕业于新加坡南洋理工大学，获统计学专业理学博士学位，并在新加坡国立大学任博士后一年。2021年12月入职暨南大学经济学院统计系，研究方向为生存分析, 贝叶斯估计，半参数模型，机器学习。论文曾发表于Computational Statistics and Data Analysis, Journal of Statistical Planning and Inference 等国际期刊。

时间：2023年7月14日（周五）上午9:30-11:30

地址：腾讯会议：387680301

邀请人: 胡涛