What FluCast Does
FluCast is a real-time epidemiological forecasting platform that ingests the latest weekly CDC FluView/ILINet surveillance data, retrains its prediction pipeline, and projects upcoming influenza-like illness (ILI) activity across configurable horizons.
Data Sources
- • CDC FluView weekly influenza surveillance
- • ILINet outpatient ILI percentage
- • Delphi Epidata FluView API (programmatic access)
- • Optional NOAA weather (temperature, precipitation)
ILI is defined by CDC as fever (≥100°F) plus cough and/or sore throat. Weekly data are preliminary and may be revised.
Model
Primary: Ridge Regression with seasonal & lag features.
Optional: Random Forest, XGBoost, SARIMA for comparison.
Features: ILI_lag1–3, rolling 3/5-week averages, sin/cos week encoding, temperature, precipitation.
Refresh: Daily CDC sync · auto-retrain on new MMWR week.
Limitations
- • Forecasts assume historical seasonality patterns persist.
- • Novel variants or behavioral shifts may degrade accuracy.
- • CDC data revisions can change recent weeks retroactively.
- • National model — regional dynamics may differ.
Disclaimer
This application is for academic and public health analytics purposes only. Forecasts are estimates based on available surveillance data and should not replace official CDC guidance or professional epidemiological judgment.