Reconstructing Transmission Trees in Healthcare Setting using Bayesian Inference

Abstract

Outbreaks of multidrug-resistant bacteria, such as Klebsiella oxytoca, present critical challenges to healthcare systems worldwide. Such bacteria can cause severe infections in immune-suppressed patients, spreading through contact with infected individuals, equipment, or contaminated environments. This thesis focuses on reconstructing transmission trees in healthcare systems using Bayesian modeling, focusing on the significance of data integration for effective infection control strategies. First, the study examines how patient, and contact data generated in hospitals contribute to understanding transmission trees. Second, it explores the incremental impact on inferred transmission trees’ accuracy by incorporating different data sources, such as temporal, contact, diagnostic, and genetic data. Lastly, the study evaluates the effects of varying sampling on transmission inference accuracy. The results indicate that integrating temporal, contact, reporting, and genetic data enhances the accuracy of transmission tree reconstructions. Furthermore, our investigation into the impact of sampling revealed that increased sampling improves accuracy and reduces variability in transmission tree structure. Overall, this research emphasizes the importance of comprehensive data integration for effective infection control strategies and provides insights for managing outbreaks of multidrug-resistant organisms in hospital environments.