Generalizability of machine learning in predicting antimicrobial resistance in E. coli: a multi-country case study in Africa

Antimicrobial resistance (AMR) is a global health threat, particularly in low- and middle-income countries (LMICs). Machine learning models, trained on whole-genome sequencing data from England, can predict AMR in E. coli, targeting key antibiotics like ciprofloxacin, ampicillin, and cefotaxime. Validation of these models using an independent dataset from Africa showed varied performance across antibiotics. The Support Vector Machine was most effective in predicting ciprofloxacin resistance, while Logistic Regression showed high accuracy for ampicillin. Key mutations associated with AMR were identified for these antibiotics.