Extended Joint Models for Longitudinal and Time-to-event Data: with applications in cardiothoracic surgery

In this thesis, we developed extensions for the joint modeling framework for longitudinal and time-to-event data, motivated by various clinical research questions in cardiothoracic surgery. These extensions focus in the handling of intermediate events during follow-up, feature selection in multivariate settings such as multiple longitudinal outcomes and multi-state processes using Bayesian shrinkage priors and sensitivity analysis for missing data under the joint modeling framework.