Comprehensive modelling of pharmaceutical solvation energy in different solvents

Abstract

Crystallization is an important processing step in production of active pharmaceutical ingredients (API). This process is used to recover/separate the synthesized APIs for further processing to final solid dosage oral formulations. Control and understanding of crystallization are of great importance for development of high-efficient processing step in pharmaceutical manufacturing. Different types of models have been developed so far for prediction of crystallization, however development of a comprehensive hybrid model is lacking. In this work, a comprehensive hybrid model is developed based on quantum chemical calculations and artificial intelligence model. The quantum chemical calculation was used to compute the solvation energy of several solutes in various organic solvents. 42 data set was collected from literature, and the solvation energy was computed using Material Studios software. The calculated solvation energy was used to build a predictive model based on artificial intelligence (AI). Artificial neural network (ANN) was employed for prediction of API solvation energy in various solvents. KFold technique was used for validation of the model in which 1/3 of the data was used for validation of the trained neural network. It was indicated that the developed ANN is capable of predicting solvation energy versus solute and solvent, with high accuracy. R2 more than 0.999 was obtained for both training and validation stages, revealing the robustness of the developed ANN in predicting solvation energy of APIs. The results can be used as predictive tool for understanding and optimization of crystallization process.