Identification of Stochastic Multicompartmental Models in Tracer Kinetics Experiments

Two efficient algorithms are described for the estimation of dynamic parameters in a tracer kinetic experiment, when a stochastic multicompartmental model is utilized. The first approach is based on the sensitivity method, where the error between the model output and the system output is minimized. Sensitivity functions are calculated and model outputs are simulated by the algorithm at each iteration in order to estimate the optimal values of the model parameters. The concept of a whitening filter is used in the second method to convert the colored output error into a white sequence: the algorithm minimizes the estimated variance of the filtered sequence. Tests performed using simulated as well as real data confirm the effectiveness of both the techniques.