Identifying and Forecasting Importation and Asymptomatic Spreaders of Multi-drug Resistant Organisms in Hospital...

Area of Science:

• Computational epidemiology
• Infectious disease modeling
• Health informatics

Background:

• Healthcare-associated infections (HAIs) caused by multi-drug resistant organisms (MDROs) present a major healthcare challenge.
• Distinguishing between imported and nosocomial infections is difficult due to asymptomatic cases and diagnostic limitations.
• Existing predictive models often fail to integrate electronic health record (EHR) data or lack mechanistic understanding.

Purpose of the Study:

• To develop a novel framework, NeurABM, integrating neural networks and agent-based models (ABM).
• To leverage EHR data and mechanistic insights for improved prediction of infection importation and nosocomial spread.
• To enhance the accuracy of identifying patients at risk for MDRO infections.

Main Methods:

• Developed NeurABM, a hybrid framework combining neural networks and agent-based modeling.
• Utilized neural networks for patient-level importation prediction.
• Employed agent-based models for simulating infection transmission and identification.

Main Results:

• NeurABM significantly outperforms existing methods in identifying importation cases.
• The framework demonstrates superior accuracy in forecasting future nosocomial infections.
• Successfully integrated EHR data with mechanistic modeling for enhanced prediction.

Conclusions:

• NeurABM represents a breakthrough in accurately identifying and forecasting healthcare-associated infections.
• The integrated approach offers a more robust solution compared to standalone modeling or machine learning techniques.
• This framework has the potential to significantly improve clinical practice for infection control.