Artificial neural networks: revolutionizing computational chemistry for physicochemical property prediction

Machine learning (ML) techniques have increasingly been recognized as valuable tools in drug development, enabling more efficient and accurate predictions of physicochemical properties and biological activities of chemical compounds. The use of ML algorithms combined with topological indices allows for a more streamlined and precise analysis of molecular structures, accelerating the drug discovery process while reducing costs and experimental burden. In this study, we investigated the molecular characteristics of drugs used in the treatment of Parkinson’s disease. Each model is evaluated using standard regression metrics, including mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE), and coefficient of determination (R2), to assess the accuracy and reliability of their predictions. Obtained results highlight the effectiveness of machine learning models in providing fast, reliable, and cost-effective predictions for physicochemical properties of compounds, demonstrating their significant potential for enhancing the drug development pipeline, particularly in the context of Parkinson disease treatment. This study further establishes the applicability of machine learning algorithms as powerful tools for quantitative structure–activity relationship analysis, supporting pharmaceutical innovation and enabling more informed decision-making in drug design.