2.1 Empirical Approximation
In this approach an empirical model, chosen by the user, is fitted to the pseudo-data. This approach does not imply or require any particular structure of the full-order model. It is, however, limited to situations in which there are a small number of response variables of interest (e.g. just mean arterial blood pressure and heart rate).
There have been several applications of empirical approximations. Ooi et al. [20] used logistic and polynomial functions to approximate the relationship between warfarin exposure and response in terms of International Normalised Ratio (INR) to predict warfarin dose requirements. Gulati et al. [48] used a similar approach to derive an empirical function that approximated the relationship between the concentration of activating agents and a proposed clotting time test based on a QSP model of the coagulation network. The function was used to optimise the design of a pilot study aimed at evaluating the proposed test as a tool for monitoring of enoxaparin therapy [48]. Similarly, Dumont et al. [49] used PBPK simulated pseudo-data to develop a simplified pharmacokinetic model that was used to design a paediatric clinical trial.