2026-10-12T19:30:27Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33752332 2026-09-14T17:42:23Z category_24208 category_28849 category_27100 item_type_3 month_year_09_2026

Health Claims Corpus (HCC) v1.0: 55.5 Million Automatically Extracted Disease–Factor Records, 2010–2026, with Linked PubMed Metadata Fliar BI (24896251) Bioinformatics and computational biology not elsewhere classified Natural language processing Health informatics and information systems biomedical text mining automated annotations disease-factor relations research dataset PubMed metadata disease-factor extraction corpus evaluation <p>The Health Claims Corpus (HCC) v1.0 contains 55,545,265 automatically extracted disease–factor records linked to 25,981,473 source document identifiers. A row represents an automated extraction record, not an independently verified clinical finding.</p><p>The release includes five linked datasets: extraction records, source-document metadata, disease–factor pair summaries, disease-term summaries and factor-term summaries. Variables cover assigned relation and direction labels, epistemic stance, inferred study categories, assigned evidence categories and matched publication metadata. Dated records span 2010–2026, with undated records supplied separately; 2026 is incomplete.</p><p>Potential research uses include extraction evaluation, terminology normalization, exploratory networks and descriptive analysis of the extracted corpus. Researchers should validate labels and account for changing source coverage before drawing substantive conclusions. The data do not measure population disease incidence or establish causal effects.</p><p>Important limitations: automated labels have not been independently validated; disease and factor terms are free text and can contain extraction errors; matched PubMed metadata does not verify individual extraction records; sentence offsets and source quotations are not supplied. The sample is illustrative, not statistically representative. Read VALIDATION_STATUS.md and the codebook before analysis.</p><p>Produced by <a href="https://fliarbi.com">Fliar BI</a>. This is the same HCC v1.0 release submitted to Harvard Dataverse under reserved identifier 10.7910/DVN/CICSTQ; that submission was under review when this package was prepared. Repository deposit does not constitute peer review.</p> 2026-09-14T17:42:23Z Dataset Dataset 10.6084/m9.figshare.33752332.v1 https://figshare.com/articles/dataset/Health_Claims_Corpus_HCC_v1_0_55_5_Million_Automatically_Extracted_Disease_Factor_Records_2010_2026_with_Linked_PubMed_Metadata/33752332 CC BY 4.0