Back

Multi-season evaluation and analysis of categorical trend forecasts of influenza hospital admissions in the United States

Davis, J. T.; Kaur, G.; Hines, A.; Ben-Nun, M.; Venkatramanan, S.; Brooks, L.; Mathis, S.; Ajelli, M.; Litvinova, M.; Kummer, A. G.; Ventura, P. C.; Mhade, S.; Weber, D.; Shemetov, D.; DeFries, N.; McDonald, D. J.; Yamana, T.; Zepeda-Tello, R.; Shaman, J.; Yaari, R.; Pei, S.; Webber, A.; Shandross, L.; Ray, E.; Wadsworth, S.; Niemi, J.; Redman, W. T.; Mullany, L.; Posner, R.; Mallela, A.; Lin, Y. T.; Hlavacek, W. S.; Smart, A.; Gill, A. A.; Drennan, A.; Fiebiger, B. J.; Miller, E. F.; Lee, J.; Mihaljevic, J. R.; Geist, K. A.; Baltz, M.; Bernik, O.; Truong, Y.-M. B.; Chen, Y.; Grosvenor, C. J.;

2026-09-02 epidemiology
10.64898/2026.08.31.26361843 medRxiv
Show abstract

Forecasting influenza hospitalizations informs public health preparedness, yet questions remain about which types of forecasts best guide action. We evaluate categorical trend forecasts, which communicate probabilities of upcoming increases or decreases in epidemic trajectories, submitted to CDC's FluSight Forecasting Challenge between Fall-2024 and Spring-2026. Teams submitted probability distributions over five categories describing direction and magnitude of week-over-week changes in laboratory-confirmed influenza hospital admissions. We assessed performance using Ranked Probability Skill Score, Brier Skill Score, and measures of forecast-observation agreement. Most models outperformed an equal-probability baseline; the FluSight ensemble ranked among the top three in the 2024-25 and 2025-26 seasons. Forecasts were most accurate during stable periods and least during periods of rapid change, with most models underestimating observed trends. Conclusions were robust to choice of scoring metric and reference model. These results support categorical trend ensembles as an approach to communicating infectious disease forecasts that may inform public health decision-making.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.