Grey Statistics Preservation: MHLW 2040 Nursing Workforce Committee, 4th Meeting (2026-07-23) — Verified Capture and Semantic Extraction of Secondary-Medical-Area Nurse Statistics

This deposit preserves nine document artifacts from the 4th meeting of Japan's Ministry of Health, Labour and Welfare (MHLW) advisory committee on nursing workforce training and securing toward 2040 (2040年に向けた看護職員の養成・確保の在り方に関する検討会, held 2026-07-23), together with a machine-readable extraction of "grey statistics" they contain. Grey statistics are policy-critical numbers whose only public carrier is a document artifact (typically PDF), with no versioned machine-readable release, no persistent identifier, and no documented reproduction path. Slides 26–28 of 資料3 in this deposit are a live example: nurse-employment statistics at secondary-medical-area granularity for Fukushima, Kyoto, and Hiroshima prefectures, re-tabulated by the MHLW Nursing Division from municipal-level report data supplied by prefectures. These statistics exist in no official statistical product — the underlying survey (衛生行政報告例, FY2024) publishes prefecture-level figures only. Verification chain applied (greystat protocol v0.1):- P1/P2 — All artifacts fetched 2026-08-18 (JST) with HTTP headers recorded; SHA-256 computed at capture; append-only JSONL manifest.- P3 — Independent witness: Internet Archive Save Page Now triggered per URL; archived copies re-fetched via the identity (id_) endpoint; 8/8 PDFs byte-identical to local captures. The dynamic HTML committee page did not match, recorded as evidence that HTML carriers are unsuitable for bit-level identity verification.- P4 — Semantic layer: 60 rows of age-structure shares (facility type × medical area × 14 age bins) and 19 rows of per-elderly-population nurse ratios, with cell-level provenance. Quality control: row-sum check passed 60/60; double-entry cross-verification against an independent re-fetch matched 916/916 cells (script included: verify_double_entry.py). Residual extractor-level risk is documented in provenance.json.- P5 — Bitemporal record: valid time (FY2024 employment; 2025-01-01 population) distinguished from publication time (2026-07-23) and capture time (2026-08-18). Contents: raw captured artifacts (store/), append-only manifest (manifest/manifest.jsonl), extracted CSVs with provenance (data/), reference implementation (greystat.py, Python stdlib only), verification script, and documentation. Government content is redistributed under the Japanese Government Standard Terms of Use (政府標準利用規約, CC BY 4.0 compatible), with attribution to the Ministry of Health, Labour and Welfare. The extraction and tooling are by the depositor. This is worked example #1 of the grey statistics preservation protocol; the accompanying methods paper is in preparation.