Synthetic population generation with public health characteristics for spatial agent-based models.

Area of Science:

• Computational epidemiology
• Population synthesis
• Agent-based modeling

Background:

• Agent-based models (ABMs) are crucial for simulating disease dynamics and intervention impacts.
• Current ABMs often oversimplify agent population initialization, neglecting key health attitudes and spatial variations.
• Realistic population synthesis for spatial ABMs remains an underexplored area.

Purpose of the Study:

• To introduce a novel method for generating synthetic populations with realistic health attitudes and protective behaviors.
• To utilize public health surveys, rather than traditional census data, for population synthesis.
• To enhance agent-based models (ABMs) for more accurate disease transmission simulations.

Main Methods:

• Developed a novel approach to generate synthetic populations using public health surveys.
• Created two synthetic populations for individuals aged 18+ in Virginia, focusing on COVID-19 vaccine attitudes and uptake (as of Dec 2021).
• Tested the method's ability to preserve statistical relationships and spatial heterogeneity.

Main Results:

• The novel approach successfully integrated public health surveys for synthetic population generation.
• Statistical relationships between vaccine uptake, attitudes, and demographics were preserved.
• Spatial heterogeneity in attitudes and behaviors was captured at fine scales.

Conclusions:

• Integrating public health surveys into synthetic population generation enhances ABMs for disease simulation.
• This method supports more realistic exploration of population responses to interventions and potential health disparities.
• The approach has broad applicability for public health domains beyond infectious diseases, advancing data-driven decision-making.