Bayesian forecasting of disease spread with little or no local data

Area of Science:

• Wildlife disease ecology
• Epidemiology
• Bayesian statistical modeling

Background:

• Effective disease outbreak mitigation requires precise spatial information on disease occurrence and spread.
• Current management often relies on non-statistical, distance-based methods around disease detections.
• These methods may lack statistical rigor and accuracy in defining affected areas.

Purpose of the Study:

• To present and evaluate an underutilized Bayesian technique for predicting and forecasting wildlife disease spread.
• To demonstrate the application of this method using limited data from a chronic wasting disease outbreak in Michigan.
• To compare the Bayesian approach with traditional non-statistical methods.

Main Methods:

• Utilized a Bayesian statistical technique incorporating limited local data and informative priors from a previous study.
• Applied the method to predict chronic wasting disease (CWD) occurrence and spread in Michigan.
• Compared the performance of the Bayesian model against non-statistical distance-based metrics.

Main Results:

• The Bayesian technique generated statistically valid predictions of CWD occurrence and spread.
• The method proved computationally simple and required minimal local data.
• Performance evaluations showed the Bayesian approach was competitive with non-statistical metrics.

Conclusions:

• The Bayesian technique provides a robust, statistically sound framework for disease prediction and forecasting, especially in data-deficient situations.
• Its simplicity and effectiveness make it a valuable tool for wildlife disease management and other data-limited ecological systems.
• The approach allows for the incorporation of new data, offering adaptive management capabilities.