In biomedical studies, we are often interested in the association between different types of covariates and the times to disease events. Because the relationship between the covariates and event times is often complex, standard survival models that assume a linear covariate effect are inadequate. A flexible class of models for capturing complex interaction effects among types of covariates is the varying coefficient models, where the effects of a type of covariates can be modified by another type of covariates. In this paper, we study kernel-based estimation methods for varying coefficient additive hazards models. Unlike many existing kernel-based methods that use a local neighborhood of subjects for the estimation of the varying coefficient function, we propose a novel global approach that is generally more efficient. We establish theoretical properties of the proposed estimators and demonstrate their superior performance compared with existing local methods through large-scale simulation studies. To illustrate the proposed method, we provide an application to a motivating cancer genomic study.
翻译:暂无翻译