Title: Scalable joint modeling of longitudinal and competing risks time-to-event data
Authors: Gang Li - UCLA (United States) [presenting]
Abstract: Joint modeling of longitudinal and time-to-event data is useful for longitudinal data analysis with possibly nonignorable missing data and for survival analysis with time-dependent covariates that are intermittently measured and/or with measurement errors. However, current estimation and inference methods for joint models are well known to be computationally complex and costly, which do not scale well even to moderate sample size data. The aim is to improve the computational performance of joint modeling methods by developing novel techniques to exploit some specific structures in fitting a joint model. Numerical simulation results and real data illustrations will be presented.