New features for population design evaluation and optimization with R functions: PFIM Interface 3.1 and PFIM 3.2
Caroline Bazzoli (1), Sylvie Retout (1), Emanuelle Comets (1), Anne Dubois (1), Hervé Le Nagard (1), France Mentré (1)
(1) INSERM U738 and Université Paris Diderot, Paris, France
Objectives: To extend the graphical user interface version of PFIM for multiple response models and the R scripts version of PFIM to accommodate more complex models with parameters quantifying the influence of discrete covariates [1, 2]. The demonstration will show the features of these new versions.
Context: The R function, PFIM, has been developed as an efficient tool for design evaluation and optimization. It is based on the expression of the Fisher information matrix for nonlinear mixed effects models. Since 2003, several releases of PFIM have been proposed. Currently, two main versions are implemented in parallel: a graphical user interface package using the R software (PFIM Interface) and a R scripts version (PFIM). The latter requires knowledge in R programming but benefits of the latest methodological developments performed in our research team. PFIM Interface 2.1 has been proposed allowing both design evaluation and optimization but only for single response. The last release PFIM 3.0 includes the ability to deal with multiple response models [3]. Some improvements were added to both versions: the possibility to write the model under analytical form or differential equations system, the use of a library of "classical" pharmacokinetic models and the availability of the Fedorov-Wynn algorithm regarding the optimization step.
Methods / Results: We first detail the new PFIM Interface version 3.1 dedicated to design evaluation and optimization for multiple response models. This version incorporates the features that were previously released in version 3.0 of PFIM. Furthermore, the library of "classical" pharmacokinetic models has been completed by the three compartment models and a library of pharmacodynamic models is now available, supporting immediate response models (alone or linked to a pharmacokinetic model) and the turnover models. PFIM Interface 3.1 can handle either a block diagonal Fisher matrix, for a quick evaluation or optimization, or the complete one.
Then, we show the R scripts version PFIM 3.2. This version includes the same features in terms of model specification and development of the expression of the Fisher information matrix as in PFIM Interface 3.1. The key new feature of PFIM 3.2 is the computation of the Fisher information matrix for models including fixed effects for the influence of discrete covariates on the parameters, and the computation of the predicted power of the Wald test for a given distribution of a discrete covariate as well as the number of subjects needed to achieve a given power.
PFIM versions and extensive documentation are freely available on the PFIM website [4].
Conclusions: These new functions, PFIM Interface 3.1 and PFIM 3.2 will be demonstrated on several examples.
References:
[1] Retout S, Mentré F. Further developments of the Fisher information matrix in nonlinear mixed effects models with evaluation in population pharmacokinetics. Journal of Biopharmaceutical Statistics, 2003; 13(2):209-27.
[2] Retout S, Comets E, Samson A, Mentré F. Design in nonlinear mixed effects models: Optimization using the Federov-Wynn algorithm and power of the Wald test for binary covariates. Statistics in Medicine, 2007; 26(28):5162-79.
[3] Bazzoli C, Retout S, Mentré F. Fisher information matrix for nonlinear mixed effects multiple response models: evaluation of the first order linearization using a pharmacokinetic/pharmacodynamics model. Statistics in Medicine, 2009; in press.
[4] http://www.pfim.biostat.fr/.
