Machine Learning and Graph Theory towards Prediction of Drugs and Diseases

Abstract

his thesis investigates the use of machine learning approaches to predict and analyze drug properties in terms of topological indices, which are important for understanding their chemical and biological characteristics. Topological indices, generated from graphical rep- resentation of chemical formation of drugs, give a measurable assessment of the molecule’s structure and are widely utilized in drug design and discovery. Firstly, we compute the topo- logical indices of various cancerous drugs including Randic, Zagreb, Nirmala, geometric quadratic and quadratic geometric indices. Afterwards, we find the physical measures like molecular weight of these drugs. Then, we conduct the correlation analysis to capture the relationship between each pair of attributes in the data. Especially, to see the linear re- lationship between molecular weight and indices. Firstly, we develop different machine learning models to check which fits the best. The performance of the model is tested using mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE) and the sum of squared errors (R2). Mainly, we have focused on (R2). Further, we show graphical representation of few of these models. This study uses advanced machine learn- ing methods to improve the accuracy and efficiency of predicting the drug property namely molecular weight, allowing for faster screening and optimization of medication candidates. These findings might be helpful for investigating cancerous drugs used in the thesis based on topological indices at a deeper level.