Over the 2025-26 season, the model developed with the Empirical Research Assistance tool best anticipated influenza-related hospital admissions in the United States.
The U.S. Centers for Disease Control and Prevention released FluSight's end-of-season review, their weekly program, from October to May, the public, private and academic team forecasts of influenza-related hospitalizations—for the current week and the next three. The CDC uses this combined forecast to announce, state by state, the expected pressure on health services.
Of the 39 eligible models of the 2025-26 season, Google's model most closely followed the actual admissions observed. It was developed with Empirical Research Assistance (ERA), an AI tool that generates optimization algorithms for different scientific fields and whose work has been published in Nature. Google indicates that the underlying technology is now accessible to trusted testers in its experimental scientific tools, and sees it as evidence that AI, combined with human expertise, can improve the prediction of epidemics.
Read the full article on the original source
Go to the source →