Google AI Tops CDC Flu Hospitalization Forecast Evaluation

During the 2025-26 flu season, a forecasting model built with Google AI ranked first at predicting flu-related hospital admissions in the CDC's end-of-season evaluation. [1] From October through May, FluSight combined weekly submissions from government, industry, and academic teams for the current week and three weeks ahead, and the CDC used the combined forecast to communicate anticipated state-level demand for medical services. [1]

The end-of-season analysis found that, among 39 eligible models, Google's forecast most closely matched the flu season's observed hospital admissions. [1] Google developed the forecasts with Empirical Research Assistance, an AI tool that generates optimization algorithms across scientific fields; ERA research was recently published in Nature, and its underlying technology is available to trusted testers through Google's experimental science tools. [1]

The result gives Google an externally judged performance point: its forecast led a CDC field of 39 eligible models rather than only a company-run test. FluSight’s design also puts that result in context, because the CDC combines submissions from government, industry, and academic teams when communicating expected demand. ERA’s role connects the forecasting outcome to Google’s broader effort to use AI-generated optimization algorithms across scientific fields, while access to the underlying technology currently goes through trusted testing. What the sources don't address: whether Google’s model will retain its lead in future seasons or how much ERA contributed relative to other forecasting components.