Estimating the Relative Risks of Spatial Clusters Using a Predictor-Corrector Method.

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems (GIS)

Background:

• Spatial scan statistics are crucial for disease surveillance and pattern identification.
• Predicting future spatial disease clusters remains a significant challenge in public health.
• Accurate prediction aids in resource allocation and informed decision-making.

Purpose of the Study:

• To develop and evaluate a predictive model for estimating relative risks of spatial disease clusters in subsequent time intervals.
• To enhance disease surveillance and pandemic preparedness through accurate spatial risk prediction.

Main Methods:

• Proposed a predictive Markov chain model with an embedded corrector component (multiple linear regression or exponential smoothing).
• Calculated relative risks of COVID-19 mortality spatial clusters in the U.S. across seven time intervals (May 2020–March 2023).
• Iteratively predicted relative risks for top 25 clusters from intervals three to seven, assessing predictive accuracy.

Main Results:

• The proposed model demonstrated moderate to high predictive accuracies for spatial cluster relative risks.
• The corrector component effectively selected between multiple linear regression and exponential smoothing for improved predictions.
• The method shows promise for real-world application in disease analytics.

Conclusions:

• The predictive Markov chain model offers a valuable tool for forecasting spatial disease cluster dynamics.
• This approach can significantly improve public health surveillance and preparedness for future pandemics.
• Further research can refine the model for broader epidemiological applications.