Integrating Machine Learning Approaches and Topological Indices for Prediction of Electronic Properties in Anticancer Drugs

Abstract

In this study, machine learning was used to predict electronic properties of anticancer drugs using molecular descriptors. Four machine learning models were employed to understand and predict the relationship between molecular structure and properties of the molecules. The results showed that extreme gradient boosting (XGBoost) gave good prediction performance in handling complex and nonlinear patterns in the data. In all cases, the feature importance analysis showed a consistent ranking. It shows how molecular descriptors had an influence on property prediction. This study shows how molecular descriptors and machine learning may be used for predicting properties and computational drug design.