RfN, R for Nonmem: A graphical interface for Nonmem outputs
Saik Urien
Unité de Recherche Clinique Paris Centre, Hopital Tarnier, 89 rue d’Assas, 75006 Paris, France
RfN is a set of R programs aimed to rapidly visualize the goodness-of-fit that results from a NONMEM run. RfN directly manages outputs produced by Wings for Nonmem (WfN). R is a GNU program that runs under DOS, Windows, Linux, Unix operating systems (The R Project for Statistical Computing, http://www.r-project.org/). RfN has a predefined default behaviour that may be easily customised. All of the programs require the reading of NONMEM table file(s) (i.e., files created by the $TABLE record). No extensive knowledge of the R program is needed.
The main features of RFN include
- fast visualisation of model curve-fitting ,DV vs. PRED WRES vs. PRED etc and model curve-fitting of DV PRED (IPRED) - time courses, whole data set or ID by ID ;
- based on simulation, visual predictive checks plus additional features like distribution statistics or confidence intervals on predicted percentile curves, and visual display of NPDE, normalised prediction distribution errors ;
- various visual/statistical tools allowing the display of parameters/covariables distribution and correlations between them ;
- specific programs dedicated to Wings (WfN), bootstrap statistics, randomization test, log-likelihood profiling, shrinkage ;
- model-independent outlier diagnostics
Most of the programs allow data splitting and viewing according to categorical (co)variables present in the table items (separation of PK and PD data after the CMT item for example, etc...). The programs can be accessed via an interactive menu.
RfN is freely available via download from https://sourceforge.net/project/showfiles.php?group_id=29501&package_id=140129&release_id=538680 or from the author in case of problems.