Aims: With the development of common language standards, an interoperability framework and a library of (certified) models the DDMoRe project makes the use of model based drug development (MBDD) more feasible. These methods allow one to more easily investigate and evaluate current modeling methods as well as to develop new methods and algorithms in various real world settings. In light of this, one aim of DDMoRe was to develop new modeling methods and algorithms to improve MBDD.
Methods: Four areas of method and algorithm development were investigated: (1) clinical trial simulation (CTS), (2) model-based adaptive optimal design (MBAOD), (3) model diagnostics and model selection and (4) parameter estimation in complex models. The combination of these tools should allow for a more comprehensive approach to quantitative MBDD. Through simulations of virtual patients with relevant demographic and covariate data, CTS enables computation of the, for example, the anticipated success rate of a future trial. That trial can be designed through the use of MBAOD, to increase/optimize the information gleaned from the trial. With more information, more complex models that fit the current data well and are predictive of future outcomes can be more readily developed.
Results: For CTS, the Simulx R-package has been developed (simulx.webpopix.org), allowing one to simulate complex models for longitudinal data by interfacing the C++ MlxLibrary with R. For MBAOD, the MBAOD R-package has been developed (github.com/andrewhooker/MBAOD), and can be used to plan, simulate, evaluate and run adaptive experiments as your understanding of the system you are studying improves through intermediate analyses of the data. A wide variety of model evaluation and selection tools have been developed and implemented in a number of software packages including PsN (psn.sourceforge.net), Xpose (xpose.sourceforge.net), and npde (www.npde.biostat.fr). Finally, a number of tools for estimation in complex models has been developed and implemented in Monolix (lixoft.com). These tools and the methodolgy behind them have been investigated in a number of pharmacometrics case studies, some of which will be presented.
Conclusion: A number of new methods and algorithms have been developed and implemented in new or existing software tools. These tools can work together to improve quantitative MBDD.