Hybrid metapopulation agent-based epidemiological models for efficient insight on the individual scale: A...

Area of Science:

• Epidemiology
• Computational Biology
• Environmental Science

Background:

• Emerging infectious diseases and climate change pose significant global challenges.
• Mathematical models aid in understanding disease dynamics and interventions but can be computationally intensive and environmentally costly.
• Agent-based models (ABMs) provide detailed individual behavior insights but require high computational resources, while population-based models (PBMs) are efficient but lack granularity.

Purpose of the Study:

• To develop and evaluate hybrid epidemiological models that balance computational efficiency with detailed insights.
• To reduce the computational and environmental footprint of infectious disease modeling.
• To offer a flexible modeling framework adaptable to various disease dynamics and intervention strategies.

Main Methods:

• Proposed spatial- and temporal-hybrid models integrating ABMs and PBMs.
• Utilized ABMs for areas/times of specific interest and PBMs for broader population dynamics and external influences (e.g., commuting).
• Simulated hybrid models based on existing ABM and PBM frameworks, demonstrating computational savings.

Main Results:

• Achieved significant reductions in computational effort, up to 98%, compared to traditional ABMs.
• Maintained the required depth of information within the focus areas of interest.
• Demonstrated the feasibility of hybrid models for studying infectious disease dynamics with reduced environmental impact.

Conclusions:

• Hybrid epidemiological models offer a computationally efficient and environmentally conscious approach to infectious disease modeling.
• This approach allows for detailed insights at the individual level where necessary, while employing aggregated models elsewhere.
• Hybrid models contribute to green computing in the field of epidemiology.