A Repeated Block Perturbation Subsampling for Large-Scale Longitudinal Data

Large-scale longitudinal data are now commonly available due to technological advances and collaborative research efforts. One example is the use of mobile data collection app in healthcare studies (mHealth), which generates a large-scale longitudinal mHealth data. In this paper, we propose a repeated block perturbation subsampling algorithm for the analysis of large-scale longitudinal data based on generalized estimating equation. The proposed method simultaneously provides consistent point and variance estimators. We establish the asymptotic properties of the resulting subsampling estimators. We also examine the finite-sample performance of the proposed method with simulation studies and demonstrate its application using real mHealth data sets.