ggPMX: a toolbox to easily generate a comprehensive set of model diagnostic plots for population models
C. Bartels, B. Bieth, T. Dumortier, I. Baltcheva, S. Bhattacharya, I. Ludwig, I. Demin, A. Gassem, D. Renard
Novartis Pharma AG
Objectives:
A comprehensive and concise set of informative model diagnostic plots is primordial for the development of population PK and PKPD models. Even though the generation of any individual model diagnostic plot is fairly straightforward, generating a comprehensive set of plots adapted to a particular project takes time, tends to be reprogrammed for each new modeling activity, and tends to be done somewhat differently by different pharmacometricians. At Novartis, we aimed at standardizing the set of diagnostic plots used for modeling activities in order to reduce the overall effort required for generating such plots. For this, we developed a guidance that proposes an adequate set of diagnostics and a toolbox, called ggPMX and presented hereafter, that allows generating such diagnostics at a quality sufficient for publication and submissions.
Methods:
ggPMX is an R package, i.e., a set of functions written in the R language, which is familiar to many pharmacometricians. The key components of the package are the Reader, the Generator, the Controller and the Reporter. The Reader reads model outputs from different sources (i.e. text files containing population parameters, model predictions, individual random effects, simulations and data-related inputs like covariates) and restructures these outputs into standard formats, which can easily be processed by the Generator. The Generator contains R language code to produce the plots and is factorized into a small set of flexible key functions. A set of default plots is defined in a configuration file. The configuration file can be adapted, e.g., to have different configurations for different types of modeling activities. The user will call Generator functions via wrapper functions in the Controller to produce either all the default plots or selected plots of interest. In addition to editing the configuration, the user has different options to adapt aspects of the plots to specific requirements. Plots may be adapted by setting parameters of the wrapper function that generate the plots; there exist additional wrapper functions to change aspects of the existing default plots; and the plots are, in general, returned as ggplot objects that can be further customized using ggplot functionality. The Controller serves as a user interface in order to maintain user input data as well as to call other modules. The Reporter generates sets of graphs and tables and integrates them into an output file with annotations.
Results:
Using a simple, user-adaptable wrapper function, the toolbox can produce different model diagnostic plots (e.g. residual and EBE-based plots assessing possible trends and the shape of the distributions such as IWRES vs IPRED, VPCs, observations vs predictions, etc.). By default, the output file generated by the Reporter contains the diagnostics proposed in the Novartis internal guidance; however, these can be adapted to produce different sets of diagnostics as desired by the user, and any of the plots may be customized individually. The types of customizations include modifications of the graphical parameters and stratifications. ggPMX supports the generation of an output file (PDF or Word) containing diagnostic plots for any model with a few lines of code by the user. The package is planned to be made available to the user community on CRAN at large.
Conclusion:
The first release of ggPMX will work with Monolix outputs and produce the necessary diagnostic plots mentioned in the Novartis internal guidance. Current plans are to enhance ggPMX to support NONMEM and nlmixr outputs as well.
