Assessment of Machine Learning Algorithms as an Emerging Model for Translational Research to Predict Antimicrobial Resistance in Clinically Relevant Pathogens

Antimicrobial resistance (AMR) is a growing global concern, and traditional methods are time-consuming and expensive. Machine learning (ML) models can offer a faster and more cost-effective way to predict AMR based on bacteria’s genotypic features. A framework for AMR prediction using ML was developed, involving raw data collection, pre-processing, defining models, training, and performance metrics evaluation. The study evaluated the performance of different ML models across five clinically relevant bacterial strains and 21 antibiotics, identifying essential genotypic features contributing to accurate predictions. The developed models showed good generalizability across different datasets and bacterial strains for each antibiotic.