Spatio-temporal stochastic graph-based learning for infectious disease forecasting
Researchers have developed a new spatio-temporal stochastic graph-based model for forecasting infectious diseases. This approach integrates stochastic formulations and uncertainty approximation to predict new cases, demonstrating adaptability to varying geographical network sizes. When tested on COVID-19 data from the US and chickenpox data from Hungary, the model showed competitive weekly performance across numerous counties, though with a slight delay in representing overall epidemic progression. AI
IMPACT Introduces a novel graph-based learning approach that could improve public health response to epidemics.