Multi-season evaluation and analysis of categorical trend forecasts of influenza hospital admissions in the United States
- 1MOBS Lab, Northeastern University, Boston, MA, USA
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 CDCs 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.
Related
- Nationwide evaluation of district-informed bottom-up and direct national approaches for one-week-ahead forecasting of dengue hospital admissions in Bangladesh
- Associations of short-term exposure to multiple air pollutants with hospital admissions for ischemic stroke in Lanzhou, China
- Etiology and Severity Profiles of Acute Pancreatitis in Consecutive Admissions at a Japanese Acute-care Hospital: Comparison with the 2016 Nationwide Survey
- Natural experiment on airport activity and hospital admissions in Northern Italy
- Does Hospital Accreditation Increase Inpatient Expenditure and Improve Quality of Care? Evidence from Heart Failure Admissions in County-Level Public Hospitals in China
BibTeX
@misc{davis2026multiseason,
title = {{Multi-season evaluation and analysis of categorical trend forecasts of influenza hospital admissions in the United States}},
author = {J. T. Davis and G. Kaur and A. Hines and M. Ben-Nun and S. Venkatramanan and L. Brooks and S. Mathis and M. Ajelli and M. Litvinova and A. G. Kummer and P. C. Ventura and S. Mhade and D. Weber and D. Shemetov and N. DeFries and D. J. McDonald and T. Yamana and R. Zepeda-Tello and J. Shaman and R. Yaari and S. Pei and A. Webber and L. Shandross and E. Ray and S. Wadsworth and J. Niemi and W. T. Redman and L. Mullany and R. Posner and A. Mallela and Y. T. Lin and W. S. Hlavacek and A. Smart and A. A. Gill and A. Drennan and B. J. Fiebiger and E. F. Miller and J. Lee and J. R. Mihaljevic and K. A. Geist and M. Baltz and O. Bernik and Y.-M. B. Truong and Y. Chen and C. J. Grosvenor},
howpublished = {medRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.31.26361843},
url = {https://www.medrxiv.org/content/10.64898/2026.08.31.26361843v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Multi-season evaluation and analysis of categorical trend forecasts of influenza hospital admissions in the United States
AU - J. T. Davis
AU - G. Kaur
AU - A. Hines
AU - M. Ben-Nun
AU - S. Venkatramanan
AU - L. Brooks
AU - S. Mathis
AU - M. Ajelli
AU - M. Litvinova
AU - A. G. Kummer
AU - P. C. Ventura
AU - S. Mhade
AU - D. Weber
AU - D. Shemetov
AU - N. DeFries
AU - D. J. McDonald
AU - T. Yamana
AU - R. Zepeda-Tello
AU - J. Shaman
AU - R. Yaari
AU - S. Pei
AU - A. Webber
AU - L. Shandross
AU - E. Ray
AU - S. Wadsworth
AU - J. Niemi
AU - W. T. Redman
AU - L. Mullany
AU - R. Posner
AU - A. Mallela
AU - Y. T. Lin
AU - W. S. Hlavacek
AU - A. Smart
AU - A. A. Gill
AU - A. Drennan
AU - B. J. Fiebiger
AU - E. F. Miller
AU - J. Lee
AU - J. R. Mihaljevic
AU - K. A. Geist
AU - M. Baltz
AU - O. Bernik
AU - Y.-M. B. Truong
AU - Y. Chen
AU - C. J. Grosvenor
PY - 2026
DA - 2026/08/31
PB - medRxiv
DO - 10.64898/2026.08.31.26361843
UR - https://www.medrxiv.org/content/10.64898/2026.08.31.26361843v1
AB - 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 CDCs 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.
ER -