Some statistical principles in compartmental analysis

The author describes several stochastic approaches to the compartmental system. Many of them are consistent with the deterministic counterpart in the mean. However, many of these models show that the CV function is independent of the initial number of particles in the system or could be numerically significant even for large X/sub 0/ if there is small transition rate in the system. Extension of some of these models are important in the fields of ecology, carcinogenesis, and risk assessment. Mathematically, these models do not introduce any more complexity in the numerical techniques of nonlinear estimation than their deterministic counterparts; yet they could be extremely useful in explaining a portion of the variation system. In theory, the variance-covariance structures of stochastic models assist a compartmentalist or pharmacokineticist in identifying and classifying the system under consideration. In many cases, a stochastic approach may be able to help in a proper experimental design