SpatialEpiBench: A New Benchmark for Spatiotemporal Epidemic Forecasting
SpatialEpiBench is a benchmark designed to evaluate spatiotemporal epidemic forecasting models in realistic public-health settings. It addresses the lack of standardized evaluation by including 11 epidemic datasets with rolling evaluations and outbreak-specific metrics. The benchmark tests adjacency-informed models with epidemic priors but finds most methods underperform a simple baseline.
Key facts
- SpatialEpiBench includes 11 epidemic datasets.
- It uses standardized rolling evaluations and outbreak-specific metrics.
- Adjacency-informed forecasting models with epidemic priors are evaluated.
- Most methods underperform a simple baseline.
- The benchmark addresses the gap in standardized spatiotemporal epidemic forecasting.
- Epidemic forecasting is crucial for public health response and resource allocation.
- Current evaluations often use simple chronological train-test splits.
- Spatiotemporal methods are natural candidates for improving forecasts.
Entities
—