Settings

What Pharmendium has ingested, when, and how the forecast engine is configured. Read-only: sources refresh through their scripts or your scheduled cron.

Sources ingested

0 of 5

Older than 7 days

All sources are fresh

Most recent refresh

Never

Data sources

Every source is safe and idempotent to re-run. See README.md for the full pipeline.

AIFA live shortage feed
Never ingested
Italy's current “farmaci carenti” list: the primary source for active shortages and their reported reasons.
Refresh with npm run ingest
0
Active shortages
AIFA historical archive (2018–2025)
Never ingested
Reconstructed real shortage episodes from AIFA's dated snapshot publications. Informs category duration stats and the forecast model training.
Refresh with npm run ingest:historical
0
Archived episodes
AIFA national drug registry
Never ingested
The full authorized-medicines catalog and ATC classification, so baseline forecasts can cover drugs never on the shortage feed.
Refresh with npm run ingest:registry
0
Registered drugs
0
ATC classes
FDA drug shortage signals
Never ingested
US FDASIA shortage-reason codes, matched by active ingredient: the only source that separates an active- from inactive-ingredient (excipient) cause.
Refresh with npm run ingest:fda-shortages
0
Signals on file
News signals (GDELT + Firecrawl)
Never ingested
Opportunistic news search for manufacturers and active ingredients behind concentrated shortages. Optional corroborating evidence, off by default. Firecrawl also covers active ingredients directly, which GDELT here does not.
Refresh with npm run ingest:signals
0
Articles on file

Forecast engine

Set via FORECAST_ENGINE in .env.

Active engine

Heuristic survival (v1)

FORECAST_ENGINE=tabicl but the Python environment or script is missing (see scripts/forecast/README.md), so runs fall back to the heuristic.

Firecrawl key: Configured