AUTOMOD - POLYALGORITHM FOR AN INTEGRATED ANALYSIS OF LINEAR PHARMACOKINETIC MODELS

This paper describes a new modular Fortran computer program which allows one to determine the order of a linear compartmental model, the estimates for the initial and the best-fitting parameters as well as several pharmacokinetic constants. The modified repeated integration method is employed to assess the order of the model and the initial parameter estimates. The Hartley improved Gauss-Newton method is used to find the best-fitting parameter values. Several statistics (correlation coefficient, Kolmogorof-Smirnov test, chi-square test, runs test, asymptotic standard deviations) are included to evaluate the adequacy of fit and the accuracy of estimated parameters. This method makes it possible to define unequivocally the model and characteristic pharmacokinetic parameters describing the whole blood (plasma or serum) concentrations after intravenous or extravascular single bolus drug administration. © 1979.