This grant proposal develops methods to address the complexity of modern PKPD population studies.
AIM 1 Design will investigate population experimental design. We will develop novel methodology to increase the reliability of methods for sample size calculation, develop a novel semi-parametric approach to PK and PKPD sample size calculations, generalize currently available methods for sample size calculation and optimal experimental design to arbitrary parametric models, and develop and investigate design for multi-response multi-input studies.
AIM 2 Model Selection will investigate the selection of competing structural models as well as factors such as covariates. The algorithms we propose are based on Bayesian model selection and on reversible jump Markov Chain Monte Carlo. They apply to parametric models, as well as to novel generalized models that use flexible representations (splines) for the relationship between parameters and covariates.
AIM 3 Modeling will develop complex parametric models for: Post-Operative Pain, HIV-1 Plasma/CNS T-cells trafficking, and drug induced Haematotoxicity. We will also develop a novel class of models generalizing indirect and direct action PKPD models and capable of representing a large variety of PKPD experiments. We will develop novel general mixed effect semi-parametric models that can incorporate fixed or time-varying covariates and can be used for final analysis or to aid in the development of parametric models.
AIM 4 Software implements the methodologies we propose as self-standing computer programs as well as interfaces with the widely used computer program NONMEM.
of the research we propose consists in providing state of the art methodology to aid in the development of drugs, increasing the understanding of their PKPD and the factors that explain their action in populations and individuals. This grant proposal develops methods to address the complexity of modern pharmacokinetics and pharmacodynamics (PKPD) population studies. We propose novel methodology for optimal experimental design, novel approaches to model selection, models applying to specific scientific problems ( Post-Operative Pain, HIV-1 Plasma/CNS T-cells trafficking, drug induced Haematotoxicity), and provide novel generalized PKPD representations which can be applied to a large variety of PKPD data. We also develop publicly available software implementing our design and model selection methodologies.
Showing the most recent 10 out of 18 publications
