Introduction

Solubility is one of the most basic information in a re-crystallization process and in many cases, there are only a few grams (or even mg or μg) of an expensive pharmaceutical or fine chemical to make a large number of crystallization tests.

Aim

To develop a computational procedure for prediction of drugs solubility in any mono-solvent and temperature of interest using a minimum number of experimental data points.

Methods

For achieving this purpose, here, the available solubility data sets were collected from the recently published articles and selected a minimum data point of each dataset to train a simple model based on the well-known van't Hoff equation combined with Abraham, Hansen and Catalan parameters as variables presenting the drug-solvent interactions in the solutions. After obtaining the model parameters, the next solubility data in each dataset was predicted by extrapolation method and the accuracy of model was estimated using the computation the mean percentage deviation of the back-calculated data.

Results

The model adequately trained using a minimum data point could be used as a practical strategy for predicting the solubility of drugs in mono-solvents at different temperatures with acceptable prediction error and using minimum experimental efforts.