A0270
Title: Functional principal component analysis under informative sampling
Authors: Peijun Sang - University of Waterloo (Canada) [presenting]
Dehan Kong - University of Toronto (Canada)
Shu Yang - North Carolina State University (United States)
Abstract: Functional principal component analysis has been shown to be invaluable for revealing variation modes of longitudinal outcomes, which serve as important building blocks for forecasting and model building. Decades of research have advanced methods for functional principal component analysis, often assuming independence between the observation times and longitudinal outcomes. Yet such assumptions are fragile in real-world settings where observation times may be driven by outcome-related reasons. Rather than ignoring the informative observation time process, the observational times are explicitly modeled by a counting process dependent on time-varying prognostic factors. Identification of the mean, covariance function, and functional principal components ensue via inverse intensity weighting. The use of weighted penalized splines is proposed for estimation, and consistency and convergence rates are established for the weighted estimators. Simulation studies demonstrate that the proposed estimators are substantially more accurate than the existing ones in the presence of a correlation between the observation time process and the longitudinal outcome process. The finite-sample performance of the proposed method is further examined using the acute infection and early disease research program study.
