Skip to content

Spatiotemporal Epidemic Forecasting Bottleneck

Home / Open Questions / Spatiotemporal Epidemic Forecasting Bottleneck

Background: Epidemic forecasting models often struggle to incorporate explicit spatial interactions, typically treating data as multivariate time series without accounting for geographical propagation.

Question / Future Work: There is a need to evolve current epidemic forecasting frameworks from multivariate temporal modeling toward explicit spatiotemporal modeling to capture disease spread dynamics across connected regions.

Why It Matters: Spatial spread is the fundamental nature of epidemics; current multivariate benchmarks remain limited by their focus on temporal correlations over spatial mechanisms.

Evidence: EpiCastBench is designed for multivariate forecasting, it does not explicitly capture spatial dependencies and can be extended to spatiotemporal settings.

Metadata & Links