Trove
- Role
- Solo build
- Year
- 2026
- Type
- Healthcare AI
Highlights
- Quantifies a real reporting gap: across 228 hospital systems whose CMS and IRS charity-care filings are both period-aligned and material, the median absolute difference is 25%
- Ships a 44-variable semantic dictionary that decodes HCRIS worksheet/line/column codes into tidy DataFrames and partitioned Parquet
- Dual-surfaced by design: a web lookup for people, plus fda-analyst and hcris-analyst Claude skills that read the actual filings and PDFs at query time
- Fully reproducible from public-domain sources with two scripts, MIT-licensed, and careful about redistribution terms (CDC SVI bundled; UW ADI deliberately local-only)
trove builds open-source lookup tools, parsers, and Claude skills on top of public-domain healthcare datasets that are widely cited but rarely usable in their raw form. Each area is search-first — type a name or identifier, see a single record at a time.