SCIENCE UNDER AUTOMATION
Real-time respiratory outbreak early warning
Transfer learning was evaluated retrospectively and prospectively in the 2024–25 respiratory season, with forecasts shared with public-health decision-makers.
What this result supports
Forecast production supports preparedness; it does not execute or causally evaluate an intervention.
Evidence record: Peer-reviewed prospective surveillance forecast evaluation. Checked: 2026-10-21.
Where this evidence is used
Population health
- Evaluate predictions prospectively: The prospective season evaluation tests forecasts on outcomes unavailable when predictions were made.
- Maintain a decision-ready forecasting pipeline: Data feeds, operational integration and decision-makers remain part of the surrounding infrastructure.
- Implement the intervention across settings: People and institutions delivered the intervention; forecasting systems supply predictions.