Predicting Physicochemical Properties of Various Drugs with Machine Learning’s Algorithms

Abstract

Machine learning methods are useful in computational chemistry by providing powerful tools for predicting the properties and activities of chemical structures. Diverse drugs with such variety in chemical structure and biological activity pose significant challenges for traditional drug discovery methods. However, that working with machine learning algo- rithms we can process sets of various drugs and find patterns to uncover hidden patterns and relationships. Concerning this analysis, important descriptors are the topological in- dices that characterize various structural features of molecular graphs. These indices offer mathematical expression for the molecule topology, capturing information regarding the connection and position of the atoms within the molecule. This work aims to establish the role of machine learning together with various drugs and the topological indices in order to enhance the ability of drug discovery and development. Key words: Machine learning Algorithms; Molecular Graph; QSPR Analysis; Topological indices; Artificial Neural Networks; Random Forest; Python Algorithm; Antiviral drugs.