Population pharmacokinetics: model-free approach and nonlinear mixed-effects modelling

The work is devoted to the application and further development of modern statistical methods to study pharmacokinetics of drugs. Specifically, it deals with applications and development of repeated measures analysis, so called 'population approach' methods, in the field of pharmacokinetics. hi the first part of the thesis, a new, model-free approach is developed and tested. It introduces a model-free measure of patient's exposure to drugs, and then investigates the relationships between the exposure level and covariates using various statistical techniques. Classification tree models (CART) and regression analysis are used to study various subpopulations of interest. It is shown, via simulations, that the model-free method is capable to identify predictors of exposure in a wide range of variability in the data. The non-linear mixed effect modelling is used to confirm the results of the model-free investigation. Model-free approach is successfully applied to several drugs. Non-linear Mixed Effects population models developed for the same data agree with its results. Limits of the new method are also identified. Specifically, it does not allow the estimation of the variability: either the within-subject (intra-individual) variability in response, or between-subject (inter-individual) variability of the pharmacokinetic parameters in the population. The second part of the thesis is devoted to applications of the Non-linear Mixed Effect methodology to population pharmacokinetics and dose-response analysis. Population pharmacokinetic and dose-response models of several drugs are developed. Pharmacokinetic models allow for complete characterisation of the drug's pharmacokinetics and its relationships to safety and efficacy. The developed models are used to explore the relationships between the exposure (individual Bayes estimates) and demographic predictors of exposure, and safety and efficacy of the drug. Finally, the developed models are used in simulations to guide the design of new studies.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral

Grantor dc:publisher.institution

University of Greenwich

Year dc:date.issued

1999

Author dc:creator

• Gibiansky, Ekaterina

Advisor dc:contributor.advisor

• Rennolls, Keith

• QA Mathematics

Language dc:language

en