محمدرضا احمدي كريم اباد

Prediction of Chemical Compound Solubility Using Machine Learning for Cyclodextrin-Based Drug Delivery Systems The prediction of chemical compound solubility plays a critical role in drug design, materials engineering, and chemical industries. Traditional laboratory-based methods are costly and time-consuming, whereas combining computational chemistry and artificial intelligence offers a more efficient approach. In this study, machine learning models were employed to predict the solubility of drugs suitable for cyclodextrin carriers based on molecular features. Chemical data were extracted from reliable databases and, after preprocessing, various models were trained. In addition to conventional molecular data, quantum chemical data and methods based on quantitative structure-property relationships and Monte Carlo simulations were utilized. Quantum data provide precise information about the electronic properties and energy of molecular systems, potentially enhancing the accuracy of model predictions. The results from machine learning models were compared with cyclodextrin data, including drug release kinetics. Various models, such as linear regression, support vector machines, random forest, extreme gradient boosting, gradient boosting, k-nearest neighbors, and decision trees, were eva‎luated. The findings indicate that hybrid machine learning models exhibit high accuracy in predicting drug release kinetics and can serve as complementary tools to reduce the need for experimental studies. This approach can facilitate the optimization of industrial processes and accelerate the development of new materials. Keywords: Artificial Intelligence, Computational Chemistry, Solubility Prediction, Machine Learning, Neural Networks

سيكلودكسترين , يادگيري ماشين , شبيه سازي ملكولي

cyklodextrin , Machine Learning , molecular simulation

Mohammadreza Ahmadi karimabad

Dr. Seyed Majid Hashemianzadeh