Machine learning-based prediction of acquired antimicrobial resistance in multiple bacterial species using K-mer analysis, mutation detection, and AMR gene profiling

Abstract

Antimicrobial resistance (AMR) is a major global health concern, necessitating rapid and accurate prediction methods. This study presents a machine learning based framework for predicting AMR across multiple bacterial species using genomic sequence data. The approach integrates K-mer analysis, mutation detection, and AMR gene profling to extract key genomic features relevant to resistance classification. The proposed method achieves high predictive accuracy, identifying K-mer signatures, and AMR gene variations as significant contributors. This scalable approach enhances genomic surveillance and clinical decision-making, enabling efficient AMR detection without the need for phenotypic testing.