PSM - Population Stochastic Modelling

The R ProjectPSM

Population Stochastic Modelling

The PSM package for R provides functions for estimation of linear and non-linear mixed-effects models using stochastic differential equations. Moreover it provides functions for finding smoothed estimates of model states and for simulation. The package allows for any multivariate non-linear time-variant model to be specified, and it also handels multidimentional input, covariates, missing observations and specification of dosage regimen.


  New version 0.8-5. Read more...  
  • Mixed-effects model using stochastic differential equations
    • Linear and non-linear models
    • Multivariate response
    • Multivatiate input
    • Missing observations
    • Covariates
    • Dosage regimen for PK/PD modelling
  • Parameter estimation in mixed-effects model using maximum likelihood (FOCE approximation)
    • It is possible to use a simple pooled likelihood function by using a model without random effects
    • Random effects can be normally distributed or any transformation hereof (e.g. log-normal etc.)
  • Simulation of data based on new or estimated model
  • Smoothing estimates of model states for optimal use of data

PSM is a package for R published under the GPL 2 license. The program has been developed at Informatics and Mathematical Modelling (IMM) at the Technical University of Denmark (DTU).

PSM builds upon CTSM which has been develloped mainly for single individual modelling but also handles multiple individuals using a pooled likehood. CTSM is based on Fortran and will thus generally run faster than PSM for identical models.