BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:BSU Seminar: 'Bayesian Covariance Structure Modelling of Clinical 
 Trial Data' - Stef Baas\, University of Twente
DTSTART:20230221T140000Z
DTEND:20230221T150000Z
UID:TALK196786@talks.cam.ac.uk
CONTACT:Alison Quenault
DESCRIPTION:Latent mixed effects models can be used in clinical trials to 
 describe clustered data consisting of multiple outcome types. Observations
  in the same factor level share a random effect which\, when integrated ou
 t\, yields a positive addition to the covariance between these outcomes.In
  this research\, instead we directly analyse covariance structure models w
 here the covariance structure comes from a multi-level normal latent mixed
  effect model. This model allows for a positive as well as a negative addi
 tion to the covariance between observations in the same factor level. Next
 \, the extension of the parameter range of these covariances to negative v
 alues has other benefits in statistical estimation and testing.  We constr
 uct a novel\, efficient and general procedure to perform Bayesian analysis
  under these covariance structure models\, involving a product shifted-inv
 erse gamma prior for the covariance parameters respecting the parameter ra
 nge. Our leading example will consist of data coming from the BIO-RESORT c
 linical trial\, where we apply our results to construct a Gibbs sampler fo
 r type-II interval censored event-time data.\n
LOCATION:Large Seminar Room\, East Forvie Building\, Forvie Site\, Robinso
 n Way\, Cambridge CB2 0SR
END:VEVENT
END:VCALENDAR