addressalign-toparrow-leftarrow-rightbackbellblockcalendarcameraccwchatcheckchevron-downchevron-leftchevron-rightchevron-small-downchevron-small-leftchevron-small-rightchevron-small-upchevron-upcircle-with-crosscrosseditemptyheartfacebookfolderfullheartglobegmailgoogleimagesinstagramlinklocation-pinmagnifying-glassmailminusmoremuplabelShape 3 + Rectangle 1outlookpersonplusprice-ribbonImported LayersImported LayersImported Layersshieldstartrashtriangle-downtriangle-uptwitteruseryahoo

Predictive modelling using joint models for longitudinal and time-to-event data

As usual we'll meet in the cafe from 18:30 with the speaker starting at 19:00. This month, Dimitris Rizopoulos, Assistant Professor in Biostatistics at Erasmus University Medical Center and author of the JM and JMBayes packages will introduce us to his interesting topic of: Joint models for longitudinal and time-to-event data: What are they, and how to fit them with packages JM and JMbayes. The first half of the talk will address the theory and implementation, with the second half discussing its application in actual case studies.

The last 20 years have seen an increasing interest in the class of joint models for longitudinal and time-to-event data. These models constitute an attractive paradigm for the analysis of follow-up data that is mainly applicable in two settings: First, when focus is on a survival outcome and we wish to account for the effect of an endogenous time-dependent covariate measured with error, and second, when focus is on the longitudinal outcome and we wish to correct for nonrandom dropout. In this talk we will briefly introduce this modeling framework, and talk about the challenges in designing and developing an R package to fit them.

A longitudinal study is a follow-up study, i.e., a study in which an outcome is repeatedly measured for a sample unit. Two examples:

* Biostat: In a HIV follow-up study we are interested in the association between CD4 cell counts (which is a marker of the immune system, and is measured at specific time points during follow-up for the patients under study), and the time-to-death.

* Civil engineering: We are interested in the structural integrity of buildings. In a "follow-up" study we collect data of some structural integrity indices for a group of buildings. We are interested in using these indices to predict when a building will no longer be habitable (i.e., the time-to-event here is the time until the building is not habitable anymore).

The nonrandom dropout refers to the fact that after the event has happened (i.e., in the first example patient died or in the second example building is not habitable), we can no longer collect longitudinal measurements (i.e., CD4 cell counts or structural integrity indices). Thus, the occurrence of an event causes dropout for the longitudinal outcome. Under some conditions this is termed non-random meaning that you need to postulate a joint model for the longitudinal and event time outcomes in order to have valid inferences.

In other words, this meeting represents an unique chance to learn from the experts!

Dimitris, Chris and Øyvind

Join or login to comment.

People in this
Meetup are also in:

Sign up

Meetup members, Log in

By clicking "Sign up" or "Sign up using Facebook", you confirm that you accept our Terms of Service & Privacy Policy