IDENTIFICATION OF PARAMETERS IN COMPARTMENTAL SYSTEM WHEN THE STRUCTURE IS GIVEN BEFOREHAND.

Compartmental analysis is a method for evaluating a system composed of several compartments from dynamical observations of a tracer injected into the system. The purpose of this study is to propose a method applicable to any compartment system whose structure is known in advance. It is assumed that the tracer is a radioisotope frequently used in medicine and biology. ″The sum of the squares of the residuals weighted by the reciprocal of the observed values″ is adopted as an objective function to identify the pertinent parameters. This function is derived from the likelihood function for the observed values. The constraints based on the law of constancy of tracer mass are evaluated by the asymptotic variances based on the Fisher information matrix and the Cramer-Rao inequality. Following the numerical experiments it is demonstrated that the method is accurate and useful for the identification of a compartmental system when its structure is known in advance.