A copula-based approach for generating realistic virtual patient populations in quantitative pharmacology

Description of session (include background & scientific importance): Generation of realistic virtual patient populations (VP) represents a key step in model-informed drug development and precision dosing strategies. In this context, capturing correct dependency structures between patient properties is essential for accurate prediction of inter-individual variability in therapeutic responses. Commonly used VP generation strategies often omit or simplify correlation structures between patient covariates or parameters, or rely on direct access to individual level patient data. Copulas are a unique type of multivariate joint distribution functions that are well-suited for modelling dependency structures between patient covariates for pharmacometric simulations, as well as for capturing interdependencies between parameters in mechanism-based models. Furthermore, copulas facilitate sharing of patient properties, i.e., for specific patient populations, easily. This talk will discuss how copulas can be effectively implemented in pharmacometric simulation workflows. The primary focus of this talk is to: (1) introduce the concept of copula models; (2) demonstrate copula model development with examples for adult and paediatric patient populations; (3) show how copulas can be integrated for capturing expression of drug metabolizing enzymes and transporters in a PBPK modelling workflow, and (4) demonstrate how copula models can be used to facilitate patient data sharing. In conclusion, this talk will illustrate how the copula-based approach can be used to advance current pharmacometric strategies in model-informed drug development and precision dosing.