Identifiability in tracer experiments.

The paper deals with structural identifiability of compartmental systems via tracer experiments, i.e., with the problem of stating a priori whether a planned input-output identification experiment allows the estimation of all unknown parameters of a compartmental system. Compartmental models and tracer experiments are briefly reviewed. A formal definition of structural identifiability is given on the basis of system theory concepts. A structural identifiability test is presented and necessary conditions are given which can be tested on the compartmental diagram, without writing the symbolic expressions of the input-output relationships. Three examples are presented; the first two show how to practically employ the suggested rules, while the third one refers to a pharmacokinetic model.