It is impossible to optimize a process for a target drug product with the desired profile without a proper understanding of the interplay among the material attributes, the process parameters, and the attributes of the drug product. There is a particular need to bridge the micro- and mesoscale events that occur during this process. Here, we propose а molecular engineering methodology for the continuous cocrystallization process, based on Raman spectra measured experimentally with a probe and from quantum mechanical calculations. Using molecular dynamics simulations, the theoretical Raman spectra were calculated from first principles for local mixture structures under an external shear force at various temperatures. A proof of concept is developed to build the process design space from the computed data. We show that the determined process design space provides valuable insight for optimizing the cocrystallization process at the nanoscale, where experimental measurements are difficult and/or inapplicable. The results suggest that our method may be used to target cocrystallization processes at the molecular scale for improved pharmaceutical synthesis.
Keywords: cocrystallization; machine learning; molecular engineering; pharmaceuticals.