The Centers for Disease Control (CDC) recently shared that a Google AI-powered model achieved the highest accuracy in forecasting flu hospitalizations for the 2025-26 season. This announcement highlights the growing role of advanced technology in public health planning.
Understanding FluSight
Each year from October to May, the CDC manages FluSight, a collaborative project that gathers weekly data from various academic, industrial, and government groups. These participants provide estimates for hospital admissions across the United States, looking at both the current week and a three-week forward window. This collective intelligence helps the CDC advise states on the likely demand for medical resources.
In a recent review of 39 different models, Google’s entry aligned most closely with the actual hospitalization numbers recorded throughout the season. This success reinforces our belief that combining human expertise with AI can significantly sharpen global disease forecasting.
The Role of ERA
Our forecasts were built using Empirical Research Assistance (ERA), an AI system that creates specialized optimization algorithms for diverse scientific applications. The methodology behind ERA was recently featured in the journal Nature. We are currently offering access to ERA’s core technology to a group of trusted testers as part of our broader experimental science initiatives.
