Surveillance of animal diseases through implementation of a Bayesian spatio-temporal model: A simulation example with neurological syndromes in horses and West Nile Virus.

The study presents a Bayesian-model-based approach for monitoring and predicting disease outbreaks, specifically West Nile Virus in horses, by observing syndromes reported in the population of interest.

Objective of the Study

• The primary aim of this research investigation was to demonstrate the application of Bayesian theory towards syndromic surveillance and its significance in monitoring neurological syndromes in horses susceptible to the West Nile Virus (WNV).

Bayesian Model for Syndromic Surveillance

• The study builds on the concept of syndromic surveillance, which is a strategy of monitoring the number of syndromes reported in a particular population and comparing this with the baseline rate. By using statistical methods, researchers can draw conclusions about disease outbreaks.
• Building upon this, the researchers adapted Bayesian theory, a statistical methodology for estimating probabilities, to this surveillance concept, creating a model which took into account not just the syndrome count, but also additional factors like space and time clustering of syndromes, historical syndrome data, and disease seasonality.

Application of the Model

• Evidence was given of how the new model can track and predict outbreaks of specific diseases, like the WNV, and of non-specific diseases. The former was achieved by focusing on diseases with similar information available as WNV, while the latter dealt with diseases unrelated to WNV in terms of available data.
• Data from historical WNV outbreaks was used to construct the model, along with surveillance data for neurological syndromes in horses.
• The model was run using a Gibbs sampling procedure, a statistical technique used for approximating a multivariate probability distribution.

Outcomes and Future Implications

• The effectiveness of the Bayesian model was established in terms of sensitivity and specificity. This was done to ensure the model was correctly identifying true positive and true negative instances.
• The study’s authors argue that predictive modelling of syndromes, such as this, can contribute significantly towards decision-making processes in animal health surveillance. This can enable more timely and effective interventions during disease outbreaks.