Artificial intelligence optimization of Alendronate solubility in CO2 supercritical system: Computational modeling and predictive simulation

Abstract

Finding different technical procedures to increase the solubility of orally-taken medicines play a vital key role towards reducing their undesirable side effects and improving their therapeutic effectiveness. Low solubility of drugs may result in the emergence of disparate challenges like poor gastro-intestinal absorption, inadequate bioavailability, and difficulty of metabolism. In this research, it has been tried to model the solubility of Alendronate medication based on two input parameters of temperature and pressure. The pressure was considered to be between 120–300 bar, and temperature was set between 308–338 K for the entire analysis. For this purpose, the support vector machine model is considered. This model and the bagging and boosting models that have been used to strengthen it have been evaluated as three different models. Based on the R criterion, the SVR model has a score of 0.926, while Begging and AdaBoost have scores of 0.881 and 0.983, respectively. Based on this, the AdaBoost model can be considered a more successful ensemble model that has increased the SVR performance. Using this ensemble method, the RMSE error rate is 4.30E-02 and MAE is 2.96E-02.

Keywords

• Solubility
• Optimization
• Machine learning
• Drug analysis
• Supercritical fluid