Validating agent-based simulation model of hospital-associated Clostridioides difficile infection using primary...

Area of Science:

• Epidemiology
• Computational modeling
• Infectious disease dynamics

Background:

• Agent-based models (ABMs) are vital for simulating infectious diseases.
• Validating these models, especially for hospital settings, is critical for accurate predictions.
• Existing models often lack hospital-specific details, limiting their real-world applicability.

Purpose of the Study:

• To present an alternative validation method for hospital ABMs.
• To introduce a new metric for validating the social-environmental network structure of ABMs.
• To adapt and validate an ABM for Clostridioides difficile infection (CDI) spread in a specific academic hospital.

Main Methods:

• Adapted a generic CDI ABM to a 426-bed Midwestern academic hospital (H-ABM).
• Incorporated hospital-specific layout, agent behaviors, and primary data.
• Validated the model against observed CDI rates (2013-2018) and used colonization pressure to assess network structure.

Main Results:

• The H-ABM successfully replicated historical CDI trends, including a 46% reduction during increased infection control efforts.
• High CDI burden in social-environmental networks significantly increased infection risk (Risk Ratio: 1.37).
• The impact of infection control interventions differed between the H-ABM and the generic model.

Conclusions:

• Hospital-specific ABMs offer a valuable validation approach when large-scale calibration is unsuitable.
• The proposed metric effectively validates the socio-environmental network structure of ABMs.
• Hospital-specific modeling enhances the utility and accuracy of infectious disease simulations in healthcare settings.