Medicaid Managed-Care External Quality Review (EQR) Findings: a national open dataset (Dark Health Data)

Dark Health Data is an open-source project that turns buried public-record health documents (PDFs) into research-ready, provenance-stamped datasets via large language model extraction (Anthropic Claude claude-haiku-4-5) with a verification layer: source grounding, symbolic and neurosymbolic (Logical Neural Network-inspired) constraint checks, a heterogeneous model ensemble, and a conformal selective-acceptance gate. Source documents are U.S. state Medicaid/CHIP managed-care External Quality Review (EQR) annual technical reports (42 CFR §§438.350–438.364), which describe validated performance measures, performance improvement projects (PIPs), and compliance findings. This release (v0.3.0) contains 10,488 records extracted from 41 source documents across 32 states/jurisdictions: 5,043 quality measures, 4,251 compliance findings, 1,194 performance improvement projects. Every record carries full provenance (source-document SHA-256, page, extraction model, and confidence) and a quality/trust score; nothing is imputed or dropped — problems are flagged with QA codes and a low trust score. These records are AI-extracted. A human-validation study (field accuracy, inter-rater agreement, and calibration) is underway; a validated release will follow as a new version of this record. Treat the data as preliminary and verify values against the linked source pages. Derived entirely from public-record documents; contains no protected health information and is not human-subjects research. Code (Apache-2.0): https://github.com/sanjaybasu/dark-health-data. Data license: CC0-1.0.