Findings meet the current draft rules. Selection rules · Definitions and source coverage
Candidate selection and experimental model evaluation.
The model uses recent monthly observations and summary features to learn heuristic labels. Its score ranks existing candidates.
The test measures agreement with draft rule labels, not real-world success. It does not yet demonstrate discovery of new case definitions.
Query the same observations and findings used in this report.
/api/stories?limit=2Send a request to see the response.
This public edition serves read-only queries. Editing, reviews and retraining remain private.
These are retrospective findings from a saved snapshot. Incomplete calendar months and invalid count days are excluded.
Case-study text is generated from the computed facts. Causes are left for investigation.
Source repositories ↗