A numerical method for fitting compartmental models directly to tracer data.

A numerical method is presented for obtaining the exchange rates of a metabolic compartmental system, working directly from the time course of tracer activity and excretion after bolus injection. The conventional mathematical formulation of this problem, in terms of sums of exponential functions, is bypassed, enabling implementation of the method in a short computer program. An important by-product of the method is a set of variance-covariance estimates for the fitted parameters. The method is tested on "classic" data, both simulated and real. Its capabilities and limitations are demonstrated, and the strategy of its effective application is discussed.