Combination of the Double Log-Log Model with Abraham Solvation Parameters to Predict Solubility of Drugs in Ethanol plus Water Mixtures

Abstract

Mathematical modeling of physicochemical properties such as solubility is an interesting issue for drug discovery and development. Modeling and prediction of solubility in cosolvent + water mixtures can help in finding appropriate solvent mixtures without experiments. The solubility of drugs in ethanol + water mixtures (up to 50 % of cosolvent) is applied to develop a quantitative structure-property relationships (QSPR) model, by combination of the double log-log model and different descriptors. The prediction capability and the accuracy of the models were checked using statistical parameters and the leave-drug-out cross-validation method. The combined double log-log model with Abraham solvation parameters gives the best results (22.4 % overall mean percentage deviation). In addition, the prediction capability of the developed model was compared with previously reported models in the literature. The results confirm that the developed model in this study has an overall mean percentage deviation less than the ideal mixing (46.6 %) and log-linear (40.9 %) models. These results show that the combination of the double log-log model with computational Abraham solvation parameters is an acceptable model to predict solubility of drugs in ethanol + water mixtures up to 50 % of cosolvent ethanol.