FluSPARC (Flu Surveillance, Prediction, Analytics, and Response Center) - This project will develop the FluSPARC (Flu Surveillance, Prediction, Analytics, and Response Center) site, managed by the University of Virginia Biocomplexity Institute (UVA-BI), as a core contributor to the CDC’s national Network of Influenza Modeling and Forecast Centers. The purpose of the network is to bolster CDC’s use of influenza modeling and forecasting to enhance situational awareness; strengthen prevention and control strategies; advance the science and performance of forecasting; and improve communication with public health partners. FluSPARC will produce open-source tools, public-facing datasets, and operational modeling capabilities to support public health decision-making and data-drive responses to seasonal and pandemic influenza outbreaks. The FluSPARC team includes experts in computational epidemiology, influenza modeling, high-performance computing, statistical methods, and AI. Partners include UVA scientists at the Biocomplexity Institute, School of Data Science, and School of Engineering. The center will collaborate with data partners including Kinsa, SleepCycle, and Metaculus; with national experts in modeling and statistical methodology at Argonne National Laboratory; and with the Virginia Department of Health (VDH). It Core activities of the Center include establishing a fully coordinated network site; producing recurring forecasts and sharing monthly findings; participating in cross-network meetings; improving modeling methods; documenting lessons learned; and developing and maintaining a Data Management Plan aligned with CDC requirements. FluSPARC will generate national and subnational influenza projections using synthetic populations and agent-based models, supported by HPC pipelines designed for large-scale simulations. These models incorporate detailed digital-twin representations of U.S. populations, customizable disease states, dynamic immunity, and intervention strategies. FluSPARC will extend these models to address public health questions, integrate behavior and socioeconomic data, and give special consideration to populations with limited access to prevention and treatment measures. The team will produce weekly FluSight submissions using ensemble frameworks. Novel accuracy and robustness measures will be used to enhance evaluation of forecast accuracy. The team will also develop advanced forecasting methods, including AI, that incorporate textual and contextual signals. FluSPARC will also contribute to network-level efforts by developing methods to harmonize assumptions, validate models, and aggregate scenario projections. Pandemic preparedness will be enhanced by integrating genomic, immunological, behavioral, and epidemiological data. Additional work will develop early predictors of seasonal severity, subtype forecasting, and substate-level forecasting using genomic and local surveillance data.
