Yang Hojin, Zhu Hongtu, Ahn Mihye, Ibrahim Joseph G
Department of Statistics, Pusan National University, Busan, South Korea.
Department of Biostatistics, University of North Carolina at Chapel Hill, USA.
Stat Methods Med Res. 2021 Aug;30(8):1917-1931. doi: 10.1177/09622802211012015. Epub 2021 Jul 4.
The aim of this paper is to develop a weighted functional linear Cox regression model that accounts for the association between a failure time and a set of functional and scalar covariates. We formulate the weighted functional linear Cox regression by incorporating a comprehensive three-stage estimation procedure as a unified methodology. Specifically, the weighted functional linear Cox regression uses a functional principal component analysis to represent the functional covariates and a high-dimensional Cox regression model to capture the joint effects of both scalar and functional covariates on the failure time data. Then, we consider an uncensored probability for each subject by estimating the important parameter of a censoring distribution. Finally, we use such a weight to construct the pseudo-likelihood function and maximize it to acquire an estimator. We also show our estimation and testing procedures through simulations and an analysis of real data from the Alzheimer's Disease Neuroimaging Initiative.
本文的目的是开发一种加权函数线性Cox回归模型，该模型考虑了失效时间与一组函数和标量协变量之间的关联。我们通过纳入一个全面的三阶段估计程序作为统一方法来构建加权函数线性Cox回归。具体而言，加权函数线性Cox回归使用函数主成分分析来表示函数协变量，并使用高维Cox回归模型来捕捉标量和函数协变量对失效时间数据的联合效应。然后，我们通过估计删失分布的重要参数来考虑每个受试者的未删失概率。最后，我们使用这样的权重来构建伪似然函数并将其最大化以获得一个估计量。我们还通过模拟以及对阿尔茨海默病神经影像倡议的真实数据进行分析来展示我们的估计和检验程序。
