Kris Reports
Structured review of Kris-format pathology reports — parsing, quality checks, and downstream analytics
What this project is
Kris is the structured/synoptic report format used in several sign-out workstreams at Memorial. This page tracks the work of ingesting those reports, validating them against CAP/ICCR expectations, and turning them into analysis-ready tables.
Objectives
- Parse Kris reports into a stable schema (one row per case, one column per synoptic field).
- Flag reports with missing or inconsistent CAP-required elements.
- Produce per-organ dashboards of completeness and turnaround.
- Feed the cleaned tables into the Quality Research and Morphology tracks.
Status
| Stream | Status | Notes |
|---|---|---|
| Parser (colon) | Working prototype | Validated on ~500 anonymized reports |
| Parser (pancreas, breast) | In progress | Field dictionary under review |
| Compliance checker | Integration stage | Wired to pathology-report-checker-skill |
| Dashboard | Planned | Target: Quarto-rendered weekly HTML |
Data
- De-identified Kris XML / plain-text exports from the Memorial archive.
- Linking key: hashed accession → case identifier maintained outside the repo.
- No raw patient identifiers live in code or artifacts.
Tools and repositories
pathology-report-checker-skill— rule-based CAP/ICCR compliance checker used as the QC backend.pathology-report-colon-QA— colon-specific QA reference implementation.patolojirapor,patolojiraporlari— historical report corpora used for schema discovery.
How to contribute
- Clone
pathology-report-checker-skilland run the sample pipeline. - Pick one organ (colon is the most mature) and propose one missing synoptic field check.
- Open a PR with both the rule and a unit test built from a synthetic report.
Next steps
- Finalize pancreas field dictionary.
- Add per-organ completeness scorecard to this page.
- Publish an internal methods note once breast parser is stable.