PharmML 1.0 - An Exchange Standard for Models in Pharmacometrics
Maciej J Swat (1), Sarala Wimalaratne (1), Niels Rode Kristensen (2)
(1) EMBL-EBI, Hinxton UK, (2) Novo Nordisk A/S, Denmark
Objectives: The lack of a common standard for exchange of models between different software tools used in population pharmacokinetics/ pharmacodynamics (e.g. NONMEM, Monolix and BUGS) has been a longstanding problem in the field. PharmML is intended to become such standard. It is being developed by the DDMoRe consortium, a European Innovative Medicines Initiative (IMI) project.
Methods: PharmML has been developed based on requirements provided by the DDMoRe community, including numerous academic and EFPIA partners, in the form of use cases for various estimation and simulation tasks (encoded in languages such as NMTRAN and MLXTRAN) and documents outlining the mathematical/statistical background (, ). The standard is developed as an XML schema definition, and existing standards are reused where possible (e.g. UncertML is used to encode variability/uncertainty).
Results: The current version supports Maximum Likelihood Maximization and Bayesian methods for models used in analysis of continuous/discrete longitudinal population PK/PD data with
- Structural models defined as a system of ordinary differential equation (ODE) and/or as algebraic equations.
- A flexible parameter model allowing for implementation of arbitrary parameter type used in the majority of models with discrete or continuous covariates.
- A nested hierarchical variability model capable of expressing very complex variability structures.
- An observation model supporting untransformed or transformed continuous, categorical, count or time-to-event data.
- A trial design model, based on a CDISC standard (), allowing for definition of common designs such as parallel or crossover with virtually any administration type.
- Typical modelling steps such as estimation or simulation based on inline or externalised experimental data sets.
Conclusions: The current PharmML specification allows already for the implementation of standard pharmacometric models and is a solid base for further development of PharmML over the remaining two years of the project. Subsequent releases will support delay and stochastic differential equations, optimal experimental design, etc.
 Lavielle, M. and Inria POPIX Team (June 2013). Mixed effects models for the population approach. URL: http://popix.lixoft.net/.
 Keizer, R. and Karlsson, M. (2011). Stochastic models. Technical report, Uppsala Pharmacometrics Research Group.
 CDISC SDM-XML Technical Committee (2011). CDISC Study Design Model in XML (SDM-XML), Version 1.0. Technical report.