Key points are not available for this paper at this time.
Antimicrobial resistance (AMR) is rising worldwide, and a better understanding of the genetic and socioeconomic determinants tied to it may establish a vantage point for surveillance and intervention. Unfortunately, the interactions between antibiotics, pathogens, and their environments are complex and deeply intertwined. Here, we present a novel machine learning and forecasting approach, integrating genomics, antibiotic phenotyping, and socioeconomic and environmental variables, designed to uncover hidden correlations and trends. Through the analysis of 45,616 bacterial genomes from 16 pathogens, 298,178 resistance profiles, and 1,112 social, economic, and environmental indicators collected across 127 countries, we identified 210 pathogen-specific AMR traits projected to increase by 2050, together with the key indicators associated with these trends. These traits were identified using structure-aware mixed-effects models with cluster-grouped cross-validation, controlling for lineage dependence. The 32 most critical rising traits were strongly linked to indicators of socioeconomic disparity. These findings provide a roadmap for targeted AMR interventions.
Baker et al. (Mon,) studied this question.