Rikard Nordgren (1), Sebastian Ueckert (1), Svetlana Freiberga (1), Gunnar Yngman (1), Andrew C. Hooker (1) and Mats O. Karlsson (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Sweden
PsN is an open source toolbox for population PK/PD model building using NONMEM. It has broad functionality ranging from results extraction to advanced computer-intensive statistical methods. PsN simplifies the organization of NONMEM output files, helps with starting jobs on different types of clusters (i.e. slurm, torque, sge and lsf) and can perform a cornucopia of different statistical methods, including: benchmark – combinatoric benchmarking of different NONMEM control stream settings, bootstrap, cdd – case deletion diagnostic to look for influential individuals, frem – full random effects modelling, resmod – residual modelling for quickly assessing appropriateness of structural [1] and residual error models [2], scm – stepwise covariate model, simeval – simulation evaluation diagnostics of outliers, sir – sampling importance resampling for parameter uncertainty assessment, sse – stochastic simulation and estimation and vpc – visual predictive check.
Updates to PsN since PAGE 2017 include a new tool, qa, which can be used to assess the quality of a model. qa uses “proxy models” (linearization [3] and residual error modelling [2]) to assess and highlight potential improvements to specific parts of a model, providing insights into the structural, variability and covariate components of a model. qa can also be used to identify influential individuals and outliers. The results of qa are presented in an automatically generated report. Another new tool, transform, can automatically add or transform variability terms in the model (ETAs) to Box-Cox or t-distributions, add interoccasion variability terms (IOV) to a parameter and convert omegas to full block form. transform also includes some more trivial transformations, such as automatically removing specific individuals and removing IIV or IOV separately from a model. transform is intended both as a help for manual model building but also to facilitate automation, via scripts, of some tedious model building tasks. In addition, the vpc tool can now create visual predictive check (VPC) plots for mixture models. One VPC plot will be generated for each mixture based on the MIXEST from NONMEM, or randomization from the individual probabilities in the phm-files [4]. Furthermore a new yaml formatted file is generated by all PsN runs containing metadata such as start and finish time, NONMEM and PsN versions used, command line, R package versions used etc. This simplifies programmatic extraction and reproducibility of runs. Finally, this year, updates have been made to frem, a tool for full random effects covariate modelling.
PsN can automatically generate plots for most of the different tools by adding the -rplots option. This automatically generates documents with, for example, visual predictive checks as part of the PsN output, without the need to manually run any R script. Many of these plots use functionality in the Xpose4 R package [5]. It is possible to customize the plots or replace them entirely by using custom R templates. These templates can either be plain R or R Markdown.
PsN is freely available at https://uupharmacometrics.github.io/PsN and the userguides for the different tools can be found at https://uupharmacometrics.github.io/PsN/docs.html
References:
[1] Ibrahim M, Ueckert S, Freiberga S, Kjellsson MC, Karlsson MO, Model-based post-processing of CWRES for assessment of prediction bias, PAGE 2018
[2] Ibrahim M, Nordgren R, Kjellsson MC and Karlsson MO PAGE 26 (2017) Abstr 7276
[3] Khandelwal A, Harling K, Jonsson EN, Hooker AC and Karlsson MO, A fast methid for testing covariates in population PK/PD Models, 2011, AAPS J 13:464-472
[4] Arshad U, Chasseloup E, Nordgren R, Karlsson MO, Development of visual predictive checks accounting for multimodal parameter distributions in mixture models, PAGE 2018
[5] Jonsson EN, Karlsson MO. Xpose–an S-PLUS based population pharmacokinetic/pharmacodynamic model building aid for NONMEM. Computer Methods and Programs in Biomedicine. 58(1):51-64.
[6] Lindbom L, Ribbing J, Jonsson EN. Perl-speaks-NONMEM (PsN)–a Perl module for NONMEM related programming. Comput Methods Programs Biomed. 75(2):85-94.
[7] Lindbom L, Pihlgren P, Jonsson EN. PsN-Toolkit–a collection of computer intensive statistical methods for non-linear mixed effect modeling using NONMEM. Comput Methods Programs Biomed. 79(3):241-57.
[8] Keizer RJ, Karlsson MO, Hooker A. Modeling and Simulation Workbench for NONMEM: Tutorial on Pirana, PsN, and Xpose. CPT Pharmacometrics Syst Pharmacol. (2013) 2, e50.
Reference: PAGE 27 (2018) Abstr 8686 [www.page-meeting.org/?abstract=8686]
Poster: Software Demonstration