Disease prevention versus data privacy: using landcover maps to inform spatial epidemic models

Area of Science:

• Veterinary Epidemiology
• Geographic Information Systems (GIS)
• Infectious Disease Modeling

Background:

• Accurate epidemiological data is crucial for predicting infectious disease spread and planning interventions.
• Limited availability of granular demographic data, particularly farm-level locations, poses a challenge in some regions.
• Existing models may struggle with aggregated data, impacting the reliability of predictions and control strategies.

Purpose of the Study:

• To assess the impact of using synthetic farm location data, derived from land cover information, on the accuracy of livestock disease spread predictions.
• To evaluate the effectiveness of geographic information system (GIS) approaches in modeling infectious disease outbreaks when individual farm data is scarce.
• To determine if improved land cover data resolution enhances the prediction of epidemic size, duration, and optimal control strategies.

Main Methods:

• Employed a geographic information system (GIS) approach to generate synthetic farm locations in the UK using land cover data.
• Compared epidemiological model predictions using synthetic farm locations (derived from broad and resolved land cover data) against predictions based on true farm data.
• Simulated an outbreak of foot-and-mouth disease to test the influence of data resolution on model outcomes.

Main Results:

• Models using broadly classified land cover data to generate synthetic farm locations showed significant deviations in predictions compared to models using true data.
• Utilizing more resolved subclass land use data for synthetic farm locations resulted in moderate to highly accurate predictions of epidemic size and duration.
• Optimal vaccination and ring culling strategies were identified with reasonable accuracy when using resolved land cover data for synthetic farm locations.

Conclusions:

• A GIS-based approach using detailed land cover data can effectively substitute for unavailable individual farm-level data in livestock disease outbreak modeling.
• This methodology offers a viable solution for predictive analyses and contingency planning in regions with aggregated demographic data.
• The findings support the use of GIS for informing policy decisions and developing robust control strategies for future livestock disease emergencies.