Logo UHasselt




Logo UHasselt Universiteit Hasselt - Knowledge in action


Dimitris Rizopoulos, Geert Verbeke, Geert Molenberghs


In follow-up studies, measurements are often collected for different types of outcomes for each subject. These may include longitudinally measured responses (e.g., biomarkers or other subject parameters) and the time at which an event of particular interest occurs (e.g., death, development of a disease or dropout from the study). These outcomes are often separately analyzed; however, in many instances, a joint modeling approach is either required or may produce better insight into the mechanisms that underlie the phenomenon under study. In this workshop, four tutorials will be given that aim to introduce different types of joint modeling techniques. Emphasis is placed on three settings: (1) time-to-event analysis with time-dependent covariates measured with error; (2) longitudinal studies with nonrandom dropout; (3) multivariate longitudinal data scenarios where the aim is to study the association structure. These joint modeling approaches are presented within a unified framework that is based on the use of random effects to explain the interdependencies between the observed outcomes.

The aim of these tutorials is to help participants clarify when a joint modeling approach is required and which models should be used depending on the actual research questions to be answered. Further, it will be made clear with numerous examples how one should construct and fit an appropriate joint model, correctly interpret the obtained results, and extract additional useful information that can help communicate the results better.

Motivation for Joint Modeling & Joint Models for Longitudinal and Survival Data

Dimitris Rizopoulos

The aim of the first tutorial is to set scene and introduce the framework of joint models for longitudinal and time-to-event data. The following topics will be covered:

  • Introduction & Motivation:  Which type of research questions requires joint modeling
  • The Basic Joint Model: Definition of joint models, assumptions, estimation, comparison with time-dependent Cox model, connection with missing data
  • Association Structures: Different functional forms to link the two outcomes

Joint Models for the Longitudinal and Dropout Processes

Geert Molenberghs

The aim of the second tutorial is to introduce joint models applicable in incomplete data settings. The following topics will be covered:

  • Introduction & Motivation: The incomplete data problem and missing data mechanisms
  • Modeling frameworks: Selection, pattern mixture and shared parameter models
  • Sensitivity analysis: why is necessary and how should be performed

Joint Models for Multivariate Longitudinal Data

Geert Verbeke

The aim of the third tutorial is to introduce joint models applicable in settings where interest in the associations between multiple longitudinal responses. The following topics will be covered:

  • Introduction & Motivation: Study association structure between different longitudinal outcomes
  • Multivariate Joint Models: Full likelihood versus the pairwise approach & pseudo-likelihood

Dynamic Predictions from Joint Models

Dimitris Rizopoulos

The aim of the fourth tutorial is to introduce the concept of dynamic predictions that has direct applications in the contexts of personalized medicine and shared decision making. The following topics will be covered:

  • Introduction & Motivation: Dynamic predictions for survival and longitudinal outcomes
  • Association Structures: How predictions are influenced by the functional form
  • Interactive web apps using shiny in R