Modelling the deviations of solubilities in water-dioxane mixtures from predicted solubilities by the Jouyban-Acree model

A quantitative-structure property relationship (QSPR) was proposed for modeling deviations of solubilities of drugs in water-dioxane mixtures at various temperatures from the predicted solubilities by the Jouyban-Acree model. The QSPR employed the normalized values of the hydration energy, energy of the highest occupied molecular orbital, total energy, logarithm of partition coefficient and molar volume of drugs computed using HyperChem software. The applicability of the proposed QSPR to real solubility data was evaluated employing 34 solubility data sets in water-dioxane mixtures at various temperatures by computing the mean percentage deviation (MPD) as an accuracy criterion. The overall MPD (± SD) of the QSPR was 17.9 ± 11.4% which is an acceptable error range in the pharmaceuticals field. The corresponding MPD (± SD) for 26 solubility data sets in ethanol-ethyl acetate mixtures at various temperatures was 9.6 ± 5.8%.