Workflow P1 · From the Meta CTIS and Google Health studies
Respiratory and COVID-19 situational awareness
What leaders see: Daily, privacy-safe signals of where illness and behaviour are shifting, checked against official data.
- Who runs it
- Epidemiology, health officer, data and privacy teams, partners
- Cadence
- Daily signals; weekly briefing
- Plugs into
- Survey platformsState and national case dataVaccination registriesDashboards and APIs
The workflow you run today, with 4 AI touchpoints
- 1Survey and study dataOpt-in responses arrive dailyPrivacy
Federated analysis: answers computed on phones; only private totals leave
- 2Weighting and checksCorrect sampling biasAgent
Automated weighting and anomaly checks on every daily batch
- 3BenchmarkAgainst cases, tests and vaccination
- 4Hotspot watchWhere signals rise firstPredict
Forecast positive tests and rising areas from survey signals
- 5Partner briefingHealth officials and agenciesLLM
Draft the weekly brief with charts and caveats; epidemiologists approve
- 6Policy and outreachTesting, messaging, resources
- 7EvaluateDid it work, for whom
Highlighted steps are where AI plugs into the existing workflow. People keep every decision.
Privacy-safe signals, validated daily
- Daily opt-in survey responses
- Official cases, tests and vaccination
- Opt-in mobile study (federated)
- Mobility and behaviour indicators
- Bayesian weighting for sampling bias
- Benchmarks against official data
- Forecasts and hotspot detection
- Differential privacy and suppression thresholds
- Health officials set measures and messaging
- Partners agree which signals to trust
- Target testing and outreach early
- Tailor messages to groups that lag
- Indicators re-validated against outcomes
- Question wording refined with partners
What changes
- Signals ahead of official reporting
- Comparable indicators across regions
- No individual can be identified from what is shared
How we measure it
| KPI | Today | With GTU | Type |
|---|---|---|---|
| Lead time over official reporting | Wait for case reports | Days earlier | Planning target |
| Indicators validated against outcomes | Unvalidated survey numbers | Every published indicator | New measure |
| Individual-level data shared centrally | Often required | None: federated, with differential privacy | New measure |
| Decisions citing the weekly brief | Ad hoc | Tracked each week | Planning target |
Targets are set with your team and tested in a pilot before anything scales.
Proof from our work: The Meta/UMD COVID-19 Trends and Impact Survey covered 114 countries and was used by WHO, CDC and national agencies (PNAS); the Google Health federated study measured masking and distancing without central data (npj Digital Medicine, 2024). Global COVID-19 impact monitoring · Privacy-preserving behaviour study
Next step
Run this workflow on your data.
Start with a free 30-minute discovery call. Most teams pilot one touchpoint first.