B1493
Title: A comparison of different R routines for joint modelling
Authors: Elisabeth Bergherr - Georg-August-Univerität Göttingen (Germany)
Anja Rappl - Friedrich-Alexander Universitaet Erlangen-Nuernberg (Germany) [presenting]
Abstract: When confronted with data that simultaneously captures outcomes of longitudinal and time-to-event processes it is advisable to model this data jointly. With the rising popularity of these joint models in recent years various \texttt{R} packages were developed incorporating the current state of research to facilitate analyses. However, the packages differ in model formulation, estimation techniques and the extent of estimated parameters. For an overview of their capabilities and limitations the packages \texttt{JM}, the pioneer amongst packages for joint modelling ,and \texttt{JoineRML} allowing for multivariate joint models as well as \texttt{JMBoost}, the first package with automated variable selection, will be compared via a simulation study and a real data example. The result may serve as guidelines as to when to use which package as well as further promote the more routine application of joint models.
