Computational study and experimental validation on the solubility of drugs in supercritical solvent for assessment of nanomedicine production via green technology for enhanced drug bioavailability

A computational method was proposed in this work for prediction of drug solubility in supercritical solvent for the drug desoxycorticosterone acetate as a case study. The main focus was on the assessment of drug candidacy for nanomedicine production via green chemistry process which does not use organic solvent. We developed the models on the solubility of desoxycorticosterone acetate (DA) drug considering two inputs: temperature (Kelvin unit) and pressure (Megapascal unit). For the modeling, the only output is drug solubility in the solvent which was considered to be mole fraction. This dataset contains 45 data rows that were collected at 5 temperatures and 9 pressure levels. We employed Decision Tree (DT), Theil-Sen (TS), and Gaussian Process Regression (GPR) core models coupled with Adaboost ensemble method and EPO for model tuning. Final generated models, namely EPA-DT, EPA-TS, and EPA-GPR have R-squared scores of 0.924, 0.882, and 0.997, respectively. Based on this fact and other analysis the EPA-GPR model is selected as the best model of this work which has RMSE error rate of 1.59 x 10(-1) and MAE error of 1.16 x 10(-1).