Approximate maximum likelihood population pharmacokinetic models for the design of dosage regimens

One of the applications of pharmacokinetics and pharmacodynamics is the design of improved drug administration regimens. In its simplest form only global descriptors of the response to a bolus are used to obtain the dosing scheme leading to the desired drug plasma concentration level. For further improvement on these simple administration regimens it is necessary to have additional information on the drug kinetics in terms of either a disposition equation model or a compartmental model. Due to the high variability of individual responses to the same dosing scheme the design of drug administration regimens is often based on the average pharmacokinetic parameters of a data base of individuals. Although widely used, this approach presents some problems since the average disposition equation model and the average compartmental model are inconsistent between themselves and with the global descriptors of the average response. In this paper methods and algorithms for the estimation of alternative population pharmacokinetic models are developed and their use is illustrated using pharmacokinetic data available from the literature.