Development of a computer program, MDGP, for population pharmacokinetic analysis written using ANSI C language on wide platforms

Abstract

MDGP, a computer program for population pharmacokinetics, has been developed. This program is based on maximum likelihood estimation with poly- dimensional normal distribution errors, and variable metric methods, direction set methods, conjugate gradient methods and the polytope method are provided as minimizations of objective function. The source codes of MDGP are described using ANSI C language. MDGP is able to run on wide platforms equipped with an ANSI C compiler. Users can use algebraic, ordinary differential and Laplace-transformed equations as descriptions of a pharmacokinetic model. Partial differential equations for each model parameter are not needed. Comparison of the estimated parameters, the iteration counts until convergence and objective function value, using MDGP and NONMEM, revealed that the calculation performance of the two programs are comparable. The MDGP is thought to be useful for the analysis of various pharmacokinetic models, including population pharmacokinetic analysis.