Research Roundup: Influenza Studies
January/February 2025 | Volume 24 Number 1
Courtesy of NIAIDMicroscopic view of influenza virus, which causes flu.
How to make better influenza A vaccines
Influenza A has two main groups, with stronger immunity within each group compared to between groups. Researchers are exploring vaccines that provide broad protection. To evaluate vaccine formulation and strategies, a team of researchers, including Fogarty’s Cécile Viboud, PhD, propose a vaccine population-level target product profile (PTPP), using models to predict the impact of future vaccines. Results suggest that a broadly protective vaccine could reduce cases of both groups and even eliminate influenza with high vaccination rates. However, a vaccine targeting just one group may lead to a resurgence of the other group if immunity is weaker than natural infection. The study also explores how a vaccine with broad, long-lasting protection could prevent a pandemic if widely adopted. The key takeaway is that future vaccines should not only be effective but also have broad coverage and long-lasting protection to better control influenza.
Evaluating influenza forecasting across two disrupted seasons
Influenza forecasting is a critical tool for public health preparedness and outbreak response. Since 2013, the CDC's FluSight challenge has engaged external research teams to submit weekly one-to-four week ahead predictions of flu activity across the U.S. The challenge originally focused on outpatient influenza-like illness (ILI) rates, but the COVID-19 pandemic disrupted this approach. By 2021, changes in outpatient care-seeking behavior and continued SARS-CoV-2 circulation had made ILI data less reliable, prompting FluSight to shift its focus to laboratory-confirmed hospital admissions—a more stable metric newly available across all U.S. jurisdictions. In the 2021-22 and 2022-23 seasons, 26 teams, including Fogarty’s Amanda Perofsky, PhD and Cécile Viboud, PhD, contributed weekly forecasts. While only about half of individual models outperformed CDC's baseline projections (which assume future flu activity matches current levels), FluSight's ensemble model—combining predictions from all teams—ranked among the top five most accurate models in both seasons. Though forecasting accuracy declined during periods of rapid change, the ensemble approach consistently provided more reliable predictions than most individual models, demonstrating the value of combining multiple strategies.
A new approach for modeling global circulation of influenza
This study proposes a new approach to better understand the global spread of seasonal influenza by combining local and international factors. Fogarty’s Nidia S. Trovão, PhD, contributed to this novel, combined approach model that integrates high-resolution demographic and mobility data, along with genetic information, to simulate flu migration across countries. The approach shows that population distribution, local mobility, and international travel, as well as seasonality are fundamental influences on influenza migration patterns. By accounting for different strains and regional behaviors, the researchers demonstrate improved predictions over simpler models. The findings suggest this method can help improve preparedness for future flu seasons and also can be applied to other epidemics, offering more accurate forecasting and better public health responses.
How adaptive ensemble models can improve influenza forecasting
Senior Author Cécile Viboud, PhD, contributed to this examination of forecasting influenza activity in tropical and subtropical regions, which have unpredictable seasonal patterns. The team developed and tested a diverse set of approaches to forecast influenza activity in Hong Kong, leveraging a multi-year surveillance record that spanned 32 epidemics from 1998 to 2019. They found that ensemble methods, which combine multiple models, significantly reduced forecasting errors. The best results came from an adaptive weight blending ensemble (AWBE), which adjusts model weights based on the most recent data, improving predictions by up to 62%. This approach proved effective in predicting influenza activity, even during irregular seasons, and could be applied to other infectious diseases or regions with similar challenges. The authors highlight the potential of combining multiple forecasting models to capture complex patterns and improve long-term predictions.
Updated February 12, 2025
To view Adobe PDF files,
download current, free accessible plug-ins from Adobe's website.
Related World Regions / Countries