Dynamic data-driven models for complex pharmaceutical reactions — the dynamic response surface methodology

Modern robotic equipment has yielded a plethora of time-resolved data collected during a set of experiments aiming to study the kinetics of a pharmaceutical reaction. This has generated the need for a modeling methodology that will represent the reaction’s time evolution. The present communication highlights the main characteristics of the Dynamic Response Surface Methodology (DRSM), which generalizes the classical Response Surface Methodology by incorporating time as an independent variable in the estimated data-driven model. We also highlight the process insights this model reveals. Besides listing the substantial number of studies that have used this type of model, we also describe how the DRSM models of all the measured species can be used to discover the stoichiometric model of a reaction system. Some comparisons with other data-driven modeling approaches are commented upon.