MIC creep in the shadows: Predictive modelling of resistance in newer antimicrobials against MDR pathogens

Antimicrobial resistance (AMR) continues to undermine the effectiveness of even our most potent, last-line antimicrobials. Newer drugs such as cefiderocol, sulbactam-durlobactam, and other β-lactam/β-lactamase inhibitor combinations are critical for treating multidrug-resistant (MDR) pathogens, particularly in ICUs. However, without proactive monitoring, their effectiveness may be compromised by emerging resistance.

Forecasting minimum inhibitory concentration (MIC) trends of newer antimicrobials can serve as an early warning system against the rise of MDR pathogens. With this vision, we plan to develop an interactive, AI-powered dashboard that predicts MIC creep in critical WHO priority pathogens, such as Acinetobacter baumannii, Pseudomonas aeruginosa, and carbapenem-resistant Enterobacteriaceae.

Our goal is to highlight global and country-level hotspots where MIC values are predicted to rise significantly, even before resistance thresholds are crossed. We believe that such predictive modelling can empower clinicians to make more informed empiric therapy decisions, strengthen antimicrobial stewardship by anticipating resistance trends, and inform public health responses with data-driven geographic and pathogen-specific insights.

To build our models, we plan to use the Pfizer-ATLAS dataset, which offers extensive global coverage with isolates spanning multiple years and regions. In addition, we aim to incorporate supplementary datasets that include MIC data for newer antimicrobials, such as sulbactam-durlobactam and cefiderocol, and region-specific datasets containing Indian isolates, to enhance local relevance and model generalizability. By focusing on the subtle but significant creep in MICs for last-line agents, we aim to create a predictive, visual, and actionable resource to support global AMR containment efforts. We believe our project will fill a critical gap in current AMR surveillance efforts and promote more targeted interventions, especially in low-resource settings.