AI-Driven Antimicrobial Discovery: Harnessing Artificial Intelligence to Combat Antimicrobial Resistance

In recent years, a large surge in antimicrobial resistance is one of the prominent problems in tackling nosocomial infections. The current arsenal of antimicrobials seems to be insufficient for treating these infections, as pathogens are evolving and manifesting various resistance mechanisms to overcome the action of existing drugs. The search for novel antimicrobial compounds is of utmost priority. However, the traditional approach of drug discovery is tedious, time-consuming and labourintensive. To address this problem, study was conducted as a computational framework that could help to screen various compounds in order to narrow down the most probable active compounds from the inactive ones. Machine Learning approach was utilized for curating a dataset from the ChEMBL database, and three machine learning classifiers (Random Forest, Support Vector Machine (SVM) and Logistic Regression) were trained to distinguish compounds based on their structural and physicochemical features. Random Forest outperformed the other two classifiers with accuracy of 95.28% and AUC-ROC of 0.986. This suggests that the model has strong ability to discriminate between antimicrobial and non-antimicrobial compounds. This study demonstrates that machine learning can be integrated into early steps of antimicrobial drug discovery to narrow down the search for novel compounds virtually from large databases.