POTS findings v1.2, package 4 of 5: long COVID, post-viral and autism intersections (26,032 findings, 1,587 papers)
收藏资源简介:
Package 4 of 5. 26,032 structured findings extracted from 1,587 papers (2020–2026) covering long COVID autonomic, autism intersection, post-viral syndromes. Findings were extracted by a language model (Anthropic claude-sonnet-5) from PubMed and Europe PMC records and full texts. The full method (docs/METHODS.md in each package), the build code and a searchable browser are listed under Related works. One block of a five-way partition of an 11,519-paper, 148,186-finding corpus of the POTS, dysautonomia and comorbid-condition literature. The five packages share no PubMed record; per-package counts sum exactly to the corpus totals. 14 studies appear as a preprint in one package and as the published version in another; annotations/paper_flags.csv links them. Self-contained. Includes the findings table (parquet and CSV), a data dictionary with fill rates measured on this slice, the QA evidence behind every stated number, the pinned extraction and validation code, and code/verify_package.py, which re-derives every headline number from the shipped data and exits non-zero on disagreement. Ontology identifiers: authority-derived (the extraction model emitted names only; a resolver looked identifiers up afterwards), scoring 0.986 on identifier-denotes-entity rate. Quote verifiability: 12.22% of rows have a provenance quote whose own source text is still available and re-checkable. Scientific fidelity is unmeasured. All QA verifies well-formedness, not whether a finding correctly interprets its source. Treat every row as a pointer to a paper to read. Paper flags: includes annotations/paper_flags.csv, which marks 2 retracted papers, 0 correction notices extracted as papers, 31 papers with a published correction and 7 duplicate records (preprint and published versions, co-publications). Exclude or count them as the file recommends; no row was removed. Checked against PubMed and Crossref on 2026-10-26. Not medical advice.



