Combined deterministic-stochastic approach for pharmacokinetic modeling

Abstract

We develop a systematic methodology for pharmacokinetic (PK) modeling to predict drug exposure profiles accurately. First, the number of compartments is determined by inspecting the singular values of an augmented matrix composed of the bolus (or impulse) drug response in the central compartment. The prediction error identification method is then applied to estimate model parameters based on the proposed deterministic-stochastic model structure in which stochastic dynamics and various dosing strategies as well as physiological delays are incorporated simultaneously. The deterministic part of the model is represented by a physical continuous-time compartmental model, whereas the stochastic part is a discrete-time empirical finite impulse response form. Three examples demonstrate that the proposed modeling strategy not only provides a good criterion to determine the appropriate compartment number but also describes well-combined deterministic-stochastic dynamics of the PK system with optimally estimated model parameters.