Model building takes a longer time than performing and analyzing simulations. Retrospective model building has two major steps—(1) data access and (2) data analysis. The former is probably the rate-limiting step. Typically, models are developed at the end of phase 3, most of the times. A prudent way to economize time to develop models is by incorporating what can be called as a "progressive model building (PMB) paradigm." The essence of the PMB paradigm is to update a model as new knowledge is accrued. The PMB is advantageous because of at least two reasons. The first one is being able to "carry-forward" the knowledge all along the drug development for a given product and the second one is being able to divide a big problem into several small components ("divide and conquer") that are easier to achieve. However, implementation of this paradigm calls for more open collaboration of scientists from all disciplines and institutional commitment to use the "current" model in designing the next trial. By utilizing the PMB paradigm, scientists are almost forced to employ mechanistic models, since the generalization power of empirical models is limited. For example, it is much easier to update the parameter estimates of an Emax model (with covariate effects) from a latest trial compared to those of a cubic-spline model.
Was this article helpful?