摘要
纵向数据和生存时间数据联合建模能减少由单独建模所引起的偏差,本文研究了基于纵向数据和生存时间联合建模的变量选择问题。对于生存时间数据,把生存时间做离散化处理,引入离散风险函数的Probit模型;对于纵向数据,利用线性混合效应模型建模。采用共享随机效应的方法对纵向数据和生存时间进行联合建模,通过利用多元高斯隐截断分布,构造出联合模型的精确似然。然后对似然函数加惩罚,重新构造目标函数,得到回归系数的稀疏估计量。理论证明以及数值模拟研究展示了稀疏估计量的良好性质。
Joint modeling of longitudinal data and survival time can the bias induced by separately modeling.This paper mainly discusses the variable selection based on the joint modeling.We discretize the survival time and introduce the Probit model of the discrete risk function,while the linear mixed effects model is used to fit longitudinal data.We adopt the shared random-effects approach to jointly modeling longitudinal data and survival time,use the multiple Gaussian hidden truncated distribution,and construct the exact likelihood of the joint model.Then,the sparse estimator of regression coefficients is obtained by reconstructing the penalized likelihood function.We investigate the performance of the proposed estimator through theoretical demonstration and a simulation study.
作者
胡亚南
王春雨
田茂再
HU Ya-nan;WANG Chun-yu;TIAN Mao-zai(Business School of Zhengzhou University,Henan Zhengzhou 45001,China;Center for Applied Statistic,Renmin University of China,Beijing 100872,China;School of Statistics,Renmin University of China,Beijing 100872,China;Xinjiang Social&Economic Statistics Research Center,Xinjiang University of Finance and Economics,Xinjiang Urumuqi 830012,China;School of Statistics and Information,Xinjiang University of Finance and Economics,Xinjiang Urumuqi 830012,China;School of Statistics,Lanzhou University of Finance and Economics,Gansu Lanzhou 730020,China)
出处
《数理统计与管理》
CSSCI
北大核心
2019年第3期483-494,共12页
Journal of Applied Statistics and Management
基金
国家自然科学基金(11861042)
中国人民大学科学研究基金项目成果(18XNL012)
全国统计科研计划项目重大项目(2016LD03)
新疆维吾尔自治区普通高等学校人文社会科学基地基金资助
关键词
纵向数据
生存数据
变量选择
联合建模
longitudinal data
survival data
variable selection
joint modeling
