A spatio-temporal statistical model to analyze COVID-19 spread in the USA

Area of Science:

• Epidemiology
• Biostatistics
• Computational Statistics

Background:

• The COVID-19 pandemic necessitates understanding disease spread dynamics.
• Spatial and temporal factors significantly influence infectious disease transmission patterns.

Purpose of the Study:

• To develop and validate a statistical model for capturing spatio-temporal dependence in COVID-19 spread.
• To assess the predictive accuracy of the proposed model for disease forecasting.

Main Methods:

• Development of a novel statistical technique employing a separable Gaussian spatio-temporal process.
• Implementation within a Bayesian framework for computational efficiency.
• Utilizing state-level COVID-19 data from the United States.

Main Results:

• A quadratic trend pattern was identified as most appropriate for modeling the data.
• Previous week's deaths were found to be a significant positive predictor of disease spread.
• The proposed model demonstrated superior predictive power compared to existing spatial and temporal models.

Conclusions:

• The developed statistical model effectively captures spatio-temporal dependencies in COVID-19 spread.
• The model offers robust short-term (1 week) and long-term (3 months) predictive capabilities.
• Findings highlight the importance of historical mortality data in predicting pandemic trajectories.