Google's AI ranks #1 for predicting flu hospitalizations.
A flu forecasting model built with Google AI ranked first in predicting hospital admissions during the 2025-26 season according to CDC evaluation results. The forecasts were generated using Empirical Research Assistance, an AI tool that creates optimization algorithms across scientific fields, with underlying technology available to trusted testers.
During the 2025‑26 influenza season a forecasting model built with Google AI achieved the top ranking for predicting flu‑related hospital admissions, according to the CDC’s end‑of‑season analysis. The evaluation covered submissions from 39 eligible models, and Google’s best entry most closely matched the observed admissions across the United States. The CDC’s FluSight initiative uses these forecasts each week to convey expected state‑level demand for medical services. The CDC’s FluSight system collects weekly predictions from government, industry and academic teams for the current week and for three weeks ahead, spanning the period from October through May. After the season concluded, the CDC compared each model’s outputs with the actual hospitalisation data to produce a ranking. Google’s model was judged on its overall alignment with the observed admissions over the entire season. The report does not disclose how the model performed for individual weeks or specific geographic regions, nor does it reveal the relative contribution of the AI tool versus human input. Uncertainty remains about how the model would fare in seasons with different strain dynamics or in countries with different healthcare reporting practices.