A Novel Method for Generating Spatially Resolved Synthetic Populations for Health Impact Assessments in Vulnerable...

Area of Science:

• Environmental Health
• Spatial Epidemiology
• Computational Social Science

Background:

• Mismatched spatial resolution between environmental exposure and sociodemographic data limits public health analyses.
• Privacy concerns restrict the availability of fine-grained population data.

Purpose of the Study:

• To develop and validate a novel algorithm for constructing address-level synthetic populations.
• To assess the impact of this synthetic population method on environmental justice (EJ) and health impact assessments (HIAs).

Main Methods:

• Developed a combinatorial optimization algorithm to create a synthetic population at the census tract level.
• Downscaled the synthetic population to the address level by matching building attributes.
• Compared address-level exposure assignment using the synthetic population versus random allocation for EJ and HIA.

Main Results:

• Significant misclassification of individual households occurred when using random address allocation compared to the matched synthetic population (21% error).
• Aggregate differences in exposure and income were modest (3.3% vs. 3.4% low-income/high-exposure households).
• Health impact assessments showed stronger differences with the synthetic population, especially with interactive effects.

Conclusions:

• Address-level synthetic populations enhance the precision and nuance of environmental justice and health impact analyses.
• The novel matching algorithm offers a privacy-preserving solution for integrating disparate data sources.
• This methodology is adaptable for broader applications in public health research across different regions and vulnerability dimensions.