Google's AI ranks #1 for predicting flu hospitalizations.

The Centers for Disease Control and Prevention announced this week that a flu forecasting model built with Google AI ranked highest in predicting flu-related hospital admissions during the 2025-26 season. The CDC’s FluSight initiative aggregates weekly forecasts from government, industry, and academic teams to project U.S. hospital admissions for the current week and three weeks ahead. These combined forecasts guide public health officials in anticipating state-level demand for medical services and planning resource allocation accordingly.

Google’s top-performing model matched observed hospital admissions more closely than 38 other eligible submissions, demonstrating the practical value of AI in public health forecasting. The model was developed using Empirical Research Assistance, an AI tool designed to generate optimization algorithms across scientific domains. This achievement underscores the potential of AI to enhance disease prediction and response strategies.

The CDC’s end-of-season analysis, published this week, evaluated 39 models submitted by various teams, with Google’s model achieving the highest accuracy in forecasting flu hospitalizations. The agency uses FluSight’s combined forecasts to communicate anticipated healthcare demand to state and local health departments, supporting preparedness efforts nationwide.

Google’s forecasting approach leverages Empirical Research Assistance, an AI system that creates optimization algorithms for scientific research. The underlying technology has been detailed in a recent publication in the journal Nature, and its experimental tools are now accessible to trusted testers as part of Google’s science-focused initiatives.