Healthcare Cost & Claim Structure (200,000 Episodes)

This upload contains the data passport and a free evaluation sample (1,000 episodes with 6,093 claim line items) — not the full dataset. The complete dataset — 200,000 episodes, about 1,180,000 claim line items, six reference dictionaries and the coherence-validation script — is available at: https://sentineldata.com.ua/dataset/healthcare-cost-claim-structure Sample contents: File (inside the sample archive) Rows Content sample_episodes.csv 1,000 Episodes EP-00000001 to EP-00001000, 32 columns, same schema as the full episodes table sample_line_items.csv 6,093 All claim line items of those 1,000 episodes, 17 columns The sample is the first 1,000 episodes by ID with all of their line items, not a stratified draw. Its mix is close to the full dataset’s nominal mix: encounter types 73.1% inpatient / 18.3% outpatient / 8.6% emergency; payers 36.5% Medicare / 32.7% Commercial / 20.6% Medicaid / 10.2% Self-pay; 931 claims paid, 48 denied and 21 partially denied; 24 claims appealed (8 overturned, 16 upheld). The reference dictionaries and validate_coherence.py are part of the full product and are not included here. We re-checked the sample against every coherence rule that can be computed from the two sample tables (bill arithmetic, DRG and diagnosis presence, adjudication logic, cost logic, date boundaries): 0 violations. Rules that need the reference dictionaries (DRG weight match, ... Full description with all tables is on the Wiki page of this project. Full dataset and purchase: https://sentineldata.com.ua/dataset/healthcare-cost-claim-structure . Zenodo copy: https://doi.org/10.5281/zenodo.23121445