A1299
Title: A generalized two-dimensional functional mixed model framework with statistical inference
Authors: Xinyue Li - City University of Hong Kong (Hong Kong) [presenting]
Abstract: Analysis of repeatedly measured functional data from dense longitudinal or spatial visits in regression models is gaining attention. While prior work has focused on continuous responses, discrete responses such as binary curves are also important. We propose a comprehensive generalized model framework to handle both Gaussian and dichotomous functional responses. The bivariate coefficient functions are estimated in a unified pointwise-smoothing pipeline using various M-estimators. We further develop inferential tools using a subject-level bootstrap algorithm to test the two-dimensional varying effect. Simulation studies compare the performance of non-robust and robust estimation and provide recommendations for choosing methods under various measurement error conditions. Applications to the Shanghai school adolescent physical activity study and the Spanish weather dataset demonstrate the utility of the proposed methods.
