Efficient modeling of drug solubility in supercritical carbon dioxide

Abstract

Development of new models and correlations to determine the drug solubility in supercritical fluids, i.e., supercritical CO2, helps us to avoid time-consuming computations with poor outcomes while improving the supercritical technology. In this communication, a supervised technique, namely, the least square support vector machine (LSSVM) was employed to estimate the solubility of 33 different drug compounds in supercritical CO2. The corresponding solubility was determined as a function of temperature, pressure, density of supercritical CO2, and two physical properties of drug, i.e. molecular weight and melting point. The results obtained from the employed LSSVM model were compared with eight correlations. The obtained average absolute relative deviation and the square of regression coefficient for the testing group of LSSVM model were found to be 5.61% and 0.9975, respectively. Therefore, the model developed in this work can be reliably applied to the studied drugs by only knowing their physical properties.