[R-pkgs] package JM -- version 0.6-0

Dimitris Rizopoulos d.rizopoulos at erasmusmc.nl
Thu Mar 18 09:01:07 CET 2010

Dear R-users,

I'd like to announce the release of the new version of package JM (soon 
available from CRAN) for the joint modelling of longitudinal and 
time-to-event data using shared parameter models. These models are 
applicable in mainly two settings. First, when focus is in the 
time-to-event outcome and we wish to account for the effect of a 
time-dependent covariate measured with error. Second, when focus is in 
the longitudinal outcome and we wish to correct for nonrandom dropout.

New features include:

* function survfitJM() has been added that calculates predictions of 
subject-specific survival probabilities given a history of longitudinal 

   * function dynC() has been added that calculates a dynamic 
concordance index for joint models. The function also returns 
time-dependent areas under the ROC curve.

   * method = "ch-GH" has been replaced by method = "spline-PH-GH" that 
fits a relative risk model with a B-spline-approximated baseline risk 
function. Similarly to method = "ch-GH", the plan is that the same 
relative risk model with a B-spline-approximated baseline risk will also 
replace method = "ch-Laplace".

   * method = "spline-PH-GH" allows for stratification, i.e., different 
spline coefficients are estimated for the different levels of a 
stratification factor. By default the knots positions are the same 
across strata -- this can be changed by either directly specifying the 
knots or by setting the control argument 'equal.strata.knots' to FALSE.

   * the new function wald.strata() can be used to test for equality of 
the spline coefficients among strata.

   * jointModel() has now the extra argument 'lag' that allows for 
lagged effects in the time-dependent covariate represented by the linear 
mixed model.

   * method = "ph-GH" that fits a relative risk with an unspecified 
baseline risk function has been renamed to method = "Cox-PH-GH".

   * a bug was corrected in joint models with piecewise constant 
baseline risk function. In particular, the 'xi' parameters were reported 
as double their actual value.

As always, any kind of feedback (e.g., questions, suggestions, 
bug-reports, etc.) is more than welcome.


Dimitris Rizopoulos
Assistant Professor
Department of Biostatistics
Erasmus University Medical Center

Address: PO Box 2040, 3000 CA Rotterdam, the Netherlands
Tel: +31/(0)10/7043478
Fax: +31/(0)10/7043014

More information about the R-packages mailing list