Utilizing phylodynamic and causal artificial intelligence methods for predicting and characterizing antibiotic resistant bacterial transmission within and beyond hospital settings - PROJECT SUMMARY/ABSTRACT Antimicrobial resistance (AMR), particularly in the context of hospital-acquired infections, poses a significant global challenge. It increases morbidity, mortality, healthcare costs, complicates infection control measures, and undermines the effectiveness of existing antimicrobial treatments. This proposal is designed to address the complexity of AMR transmission patterns, which have remained uncertain under traditional outbreak management strategies. Therefore, novel data-driven approaches are urgently needed, leveraging modern genomic and computational methods for a better understanding and control of AMR pathogens. The proposal's overarching objective is to provide a high-resolution characterization of multi-drug resistant (MDR) bacterial transmission patterns and evolution within and beyond hospitals. The project will integrate complex data sources, such as whole-genome sequencing (WGS), electronic health records (EHR) and epidemiologic data from a prospective cohort study, and then apply phylodynamic and causal artificial intelligence methodologies to achieve these goals. It focuses on two specific aims. The first will be to characterize flow of transmission from communities to hospitals in gram-negative MDR bacterial infections. The application of WGS, linked with sociodemographic and economic data, will be used to trace the lineages of these bacterial infections. Based on preliminary work on gram-positive MDR pathogens, we expect the majority of gram-negative hospital-onset infections originate from community ancestral strains rather than within the hospital setting, pointing towards independent introductions of bacterial strains into the hospital setting. The second aim will identify use EHR and causal artificial intelligence to develop a causal model that predicts WGS-confirmed MDR bacterial outbreaks with higher precision than conventional approaches. This aim will identify actionable within-hospital factors that increase the risk of MDR bacterial outbreaks. By addressing a critical gap in the interplay between hospital and community settings in the transmission of MDR bacterial strains, the outcomes of this study are expected to contribute to a potential shift in how infection control teams utilize existing data sources. Ultimately, this project will provide the trainee with the expertise in using complex data to understand the evolution of emerging pathogens and identify mechanism of transmission in hospitals. The skills cultivated from this project will enable the trainee to pursue his long-term career goal in becoming an infectious disease physician-scientist with the unique ability to integrate molecular biology, hospital epidemiology, and modern data science approaches. This research and training directly target the need for innovative data science to inform tailored interventions to combat AMR, especially among gram-negative pathogens.
