A framework for modeling county-level COVID-19 transmission

Area of Science:

• Epidemiology
• Geographic Information Systems (GIS)
• Public Health

Background:

• Understanding COVID-19 transmission is crucial for effective public health interventions.
• County-level factors like demographics, socioeconomic status, environment, and mobility influence disease spread.
• Spatial dependence and policy heterogeneity are key considerations in epidemiological modeling.

Purpose of the Study:

• To analyze COVID-19 transmission across U.S. counties using advanced spatial statistical methods.
• To compare the performance of spatial models against Ordinary Least Squares (OLS) regression.
• To investigate the impact of environmental factors and state-level policies on disease transmission.

Main Methods:

• Ordinary Least Squares (OLS) regression for baseline analysis.
• Moran's I for spatial autocorrelation detection.
• Spatial Autoregressive (SAR) and Spatial Error Models (SEM) for spatial dependence.
• Multilevel modeling for state-level policy analysis.
• Geographically Weighted Regression (GWR) for spatial non-stationarity.

Main Results:

• Spatial models (SEM) demonstrated superior fit (R²=0.6846, RMSE=1.642) compared to OLS (R²=0.4849, RMSE=2.0891).
• Significant spatial clustering of COVID-19 cases was identified.
• Environmental variables like precipitation and temperature showed localized impacts on transmission.
• State-level policies were incorporated into a multilevel framework.

Conclusions:

• Spatial modeling provides a more accurate representation of COVID-19 transmission dynamics than traditional methods.
• Geographic variations in environmental factors and policy interventions significantly influence disease spread.
• The integrated methodological framework offers a robust approach for future epidemiological studies with spatial considerations.