Tarjinder Sahota (1), Stefano Zamuner (1)
(1) Clinical Pharmacology Modelling and Simulation, GSK, UK
Objectives: Simulation from population PK/PD models enables us to quantitatively assess the clinical implications of untested pharmacological hypotheses and intervention strategies. In strategic drug development meetings, real time production of graphical outputs enables evidence based scenario planning where the breadth of clinical team expertise can be leveraged. Traditionally, this has been difficult to do due to software limitations software. The desired workflow should have:
1) Access to a fast and reliable differential equation (DE) solver
2) Functionality for mixed effects models
3) Functionality for complex dosing inputs
4) High quality graphical output with easy widget creation
5) Fast run time
The R language is widely used by PK/PD modellers. Recent package development has provided user friendly access to existing DE solver libraries. RStudio, an integrated development environment for R, also includes a user friendly package for interactivity [2]. The objective here is to illustrate the use of these packages in RStudio for real time interactive PK/PD simulations.
Methods: The following R packages were installed on a standard laptop (Intel Core i5 – 3427U @ 1.8 Ghz, 4GB RAM): deSolve [1], manipulate (installed with Rstudio), and plyr.
The “deSolve” package provides an interface to the Fortran LSODA function. The option to pre-compile enables fast solution of ODEs and an interface is provided for bolus and zero order input into compartments. “plyr” and “ggplot2” provide user friendly data manipulation and plotting capabilities. The “manipulate” package provides functionality to add sliders, checkboxes and menus to plots in Rstudio. We show example interactive output with full source code.
Results: We illustrate an interactive simulation from a population PK model with less than 60 lines of R code. We also show a simulation of a cell life cycle model with delayed differential equations. Speed of execution was predominantly limited by the ODE solver. Specification of ODEs in C++ for compilation drastically reduced run time by more than 10 fold.
Conclusions: The combined use of “deSolve” and “manipulate” packages in Rstudio enables the production of high quality, interactive graphical outputs from PK/PD simulations.
References:
[1] Soetaert, K., et al. (2010). Solving Differential Equations in R: Package deSolve Journal of Statistical Software, 33(9), 1–25.
[2] RStudio (2012). RStudio: Integrated development environment for R (v0.98.501)
Reference: PAGE 23 () Abstr 3265 [www.page-meeting.org/?abstract=3265]
Poster: Methodology - Other topics