Title: Functional survival analysis with several classes of trajectories
Authors: Yuko Araki - Shizuoka University (Japan) [presenting]
Abstract: Survival analysis with a few classes of functional covariates is considered. There are several approaches to derive the class of latent trajectories from longitudinal observation, such as latent class growth models. We first identify the class of individual trajectory by simple functional clustering method, and then we use this information as covariates in survival analysis. In a simulation study, we compare the proposed method to other latent trajectory approaches. For illustration, we use Body Mass Index data observed for a few decades in Japan.