Abstract

Joint modeling is a common approach to study association between longitudinal process and time-to-event. It has been extensively studied and multiple approaches and variations have evolved. Current approaches, however, can typically jointly analyze one or a few longitudinal processes. We propose a novel joint modeling approach that can handle high-dimensional longitudinal processes. One motivating application of this work is based on retinal images of the eye, measured longitudinally at each visit of the subject. Additionally, social determinants of health indicators of the subject can also be accessed and the goal is to develop an appropriate model for the outcome of disease progression