OPTIMAL MULTICOMPARTMENTAL EXPERIMENT DESIGN

Abstract

Compartmental modelling of biomedical systems, devoted to e.g. optimal therapy design, provides us with a compartmental system, the associated set of differential equations, and a number of unknown parameters. We estimate values of the unknown parameters so that we obtain a best fit of the model input-output behaviour to the experimental data. If the fit is not satisfactory, we remodel the structure of the system or/and reorganise the experiment design, and re-estimate the parameters. This is an iterative process, leading to the best model structure described with the most accurate values of its parameters. For satisfactory solution of the problem of optimal experiment design three co-operating computer programs are designed: MFIT, OSSP and OIN. MFIT enables to choose the best from any reasonable model functions. Reduced optimal sampling schedule (OSSP) and optimal input signal (OIN) allow to get the most accurate parameters’ value. They operate on the basis of output files of MFIT. The final optimal experiment design is obtained by repeating the procedures.