Influpaint is a forecasting tool that applies denoising diffusion probabilistic models (DDPMs) to predict seasonal influenza activity. It converts epidemic curves into images, where one axis represents time, the other location, and pixel values correspond to hospitalization or case rates. Using past observed data, the model “inpaints” the missing future section of the image through iterative denoising steps. It is trained on CDC FluView surveillance data and simulated outbreaks from the U.S. Flu Scenario Modeling Hub.