Mlxtran Documentation and Language Reference

Version 2016R1

This documentation is for Mlxtran for the Monolix Suite 2016R1



MlxTran is a declarative, human-readable language for the description of the structural and statistical elements of non-linear mixed effects model typically found in pharmacometrics modelling and simulation. MlxTran is used for data exploration with Datxplore, for model exploration with Mlxplore, for parameter estimation with Monolix and for simulation with Simulx.

Mixed effects modelling with MlxTran

The main output of every MlxTran model are the longitudinal model observations y. The observations are derived from the model prediction and the observational model (error model). The model inputs are a set of parameters (fixed effects). In the context of pharmacometric modelling the model parameters can have between patient variability and can depend on covariates.

MlxTran model types

Mlxtran can be used for the following type of models:

  • Continuous data models
  • Categorical data models
  • Count data models and
  • Time-to-event (or survival) data models.

Any combination between the different type of models is possible as well.

MlxTran model structure

Mlxtran can be used to represent models for parameter estimation with Monolix or for simulation with Simulx or Mlxplore. Because parameter estimation and simulation are different tasks, models will typically be different in some of the sections they contain. In Monolix the covariate and statistical models for the individual parameters is provided through the user interface. The design of the dosing is provided through the data set. For this reason a model for Monolix only requires the [LONGITUDINAL] section of <MODEL> as illustrated in the figure below showing the most general Mlxtran models for Monolix, Simulx and Mlxplore. Different bracket styles designate different hierarchical elements of the model.