Abstract:
|
With the advancement of technology, multiple sources and facets of data, including both physical and virtual, are collected over time. These longitudinal data are often observed intermittently, at subject-specific times with mis-matched covariates and response. Existing methods for asynchronous longitudinal data analysis stipulates the linear relationship between longitudinal covariates and response. In this paper, we propose a new method for estimating the regression coefficients in a partially linear model. The asymptotic normality of the resulting estimators is established, with a robust sandwich standard deviation formula. Simulation studies support our theoretical findings and show its favorable performance with competing methods. Dataset from an HIV study is used to illustrate our methodology.
|