Skip to contents


The lcmm package implements various extensions of mixed models. It handles continuous (Gaussian or not) and ordinal outcomes with repeated measures and latent classes. A time-to-event can jointly be considered in a proportionnal hazard model. All models are estimated with a maximum likelihood framework using a modified Marquardt-Levenberg algorithm. The package also includes several predictions, visualization, and utility functions to conduct a complete statistical analysis.

A detailed companion paper is available in Journal of Statistical Software :

Proust-Lima C, Philipps V, Liquet B. Estimation of Extended Mixed Models Using Latent Classes and Latent Processes: The R Package lcmm. Journal of Statistical Software, Articles. 2017;78(2):1-56.

And specific statistical models estimated are described in various statistical papers of the authors.


Install the package

The lcmm package needs version 3.5 or newer of the R software. To install the released CRAN version of the package, use

To get the most recent update, install it from github :


The lcmm package depends on other R package, namely :

  • survival (>=2.37-2) for dealing with the survival outcomes
  • parallel and doParallel for parallelizing some time consuming functions
  • mvtnorm for generating random parameters
  • randtoolbox for the quasi Monte Carlo sequences
  • marqLevAlg (>2.0) for the numerical optimization
  • numDeriv for computing the Hessian

To run the examples proposed in this website, the following package are also needed :

  • lattice
  • NormPsy
  • ggplot2
  • ggpubr
  • dplyr
  • splines
  • gridExtra
  • ggalluvial



This website is intended to help the lcmm users in their statistical analyses. It provides an overview of the package, several vignettes, a FAQ page and the help pages of all functions included in the lcmm package.

Further issues and questions about the use of the lcmm package are reported on the github issue page Please check both opened and closed issues to make sure that the topic has not already been treated before creating a new issue. To report a bug, please provide a reproducible example.