|Model specification in PKBugs||Graphical output in WinBUGS/PKBugs|
PKBugs is an efficient and user-friendly interface for specifying complex population pharmacokinetic/pharmacodynamic (PK/PD) models within the widely-used WinBUGS software.
WinBUGS is a Bayesian modelling framework that can be used to analyse, using Markov chain Monte Carlo (MCMC) techniques, arbitrarily complex Bayesian full probability models. Because WinBUGS is general purpose it requires a method of model specification that is suitable for any class of model. Unfortunately, generality almost always comes at the expense of efficiency, and for any given class of model, this method is unlikely to be optimal. Population PK/PD is such a class.
Indeed, although feasible, the specification of population PK/PD models for the majority of "real-life" applications, using WinBUGS in its standard form, is incredibly complicated. This is largely due to the complexity of patients' dosing histories and the fact that each dose itself typically requires a complex non-linear model. In addition, there may be further complications, such as time-varying covariates, censored observations, and outlying observations/individuals.
PKBugs alleviates these difficulties of model specification and thus makes accessible state-of-the-art MCMC techniques to practitioners in the field of population PK/PD. This is achieved via a short-hand notation for data entry (as in NONMEM (Beal & Sheiner, 1992)) along with a series of simple dialogue boxes and menu commands. PKBugs parses this model specification and constructs an object-oriented internal representation of the model that is compatible with WinBUGS. WinBUGS may then be used, in the normal way, to conduct the remainder of the analysis.
The various features and system requirements of PKBugs are outlined here.
PKBugs is FREE! Download the full version here.
A brief description of model specification in PKBugs can be found on our model specification page.
The references page lists a number of relevant articles/manuscripts.
|Last Updated December 2004|
Site maintained by:
MRC Biostatistics Unit,
Institute of Public Health,
University Forvie site,
Cambridge CB2 0SR