Datadory notebook
How Data Scientists Use Health Care REITs Data
Data scientists use health care REITs data as a two-layer problem: a thin equity panel of roughly 15-25 listed US names, joined to the government delivery statistics that drive tenant revenue. Of the 8 primary datasets Datadory catalogs for the industry, 7 are free and 5 are US-government commercial delivery terms, led by CMS's 159-dataset Data Hub.
1,744 datasets. Pick your catch.
How do data scientists actually use health care REITs data?
The workflow reality on a data science bench is inverted relative to most sectors: the listed universe is tiny - Yahoo Finance's “REIT - Healthcare Facilities” industry counts roughly 15-25 US-listed names (WELL, VTR, ARE, DOC, OHI, CTRE) - so almost none of the interesting feature engineering happens on the price tape. It happens in the delivery statistics that drive tenant revenue.
The ranked shortlist below is ordered by Datadory's quality score, as of August 2026.
What does the CMS facility layer look like in a pipeline?
Start with the macro driver. The CMS Data Hub - Medicare & Medicaid Datasets is the highest-scored record in this slice at quality 10: 159 cataloged datasets spanning Medicare, Medicaid, CHIP and Marketplace at facility, geography-and-service and national levels, for the 2013-2024 service years, delivered as bulk CSVs plus a JSON data API under public-domain terms. It is the natural base table for enrollment, spending and utilization features.
Join keys matter here: facility identifiers line up across CMS files, so a REIT property address can be matched to operator records, star ratings and downstream charge data without fuzzy matching in most cases.
Which charge series supports payer-mix features?
CMS Medicare Provider Charge Data (quality 9) is unique in this slice: charges, payments and volumes reported one row per hospital by DRG for inpatient (service years 2013-2024) and per APC for outpatient (2015-2024), across about 3,000+ IPPS and OPPS hospitals. Every hospital carries a CCN with address, ZIP, state FIPS and RUCA codes, so urban/rural and metro flags fall out of the file itself rather than a separate geo join.
Two modeling caveats. Releases are annual with roughly a one-year lag, so treat the series as a slow-moving structural covariate, not a timely signal. And coverage is Original Medicare fee-for-service only - Medicare Advantage stays never appear - so payer-mix estimates skew toward the older, FFS share of a market.
A practical transform for REIT work: aggregate DRG-level payments to hospital-by-year totals, intersect with the facility footprint from the Provider Data Catalog, and derive market-level revenue intensity around each property.
Where do demographic demand features come from?
The CDC Data.CDC.gov Open Data Portal (quality 9) is the programmatic option: an official Socrata API serving about 1,078 public-domain datasets as CSV, JSON, RDF and GeoJSON, covering chronic disease, immunization, surveillance and social determinants at national, state, county and some sub-county geographies.
Register a Socrata app token before you schedule anything - anonymous clients hit rate limits fast at that volume. And when the choice is between the CMS hub and the CDC portal as your macro table, our CMS Data Hub vs CDC Data.CDC.gov comparison lays out where each one wins.
Pick up where this leaves off
Every one of these ships with sample rows before you commit to anything.
CMS Data Hub - Medicare & Medicaid Datasets
CMS Provider Data Catalog
ccn · overall_rating · health_inspection_rating …+4 more
CMS Medicare Provider Charge Data (Inpatient & Outpatient)
CDC NCHS Data Browsing Collection
_113_cause_name · topic · topic_subgroup …+1 more
CDC Data.CDC.gov Open Data Portal
_113_cause_name · topic · topic_subgroup …+2 more
HCUP - Healthcare Cost and Utilization Project
PAY1 · HOSP_NIS · DISCWT
Want rows instead of a pitch? Name the datasets.
API, files, or your warehouse. Daily, weekly, or hourly.
Get a sampleQuestions worth asking
How do I get hospital charges by DRG and APC for modeling?
CMS Medicare Provider Charge Data publishes one row per hospital by DRG for inpatient (service years 2013-2024) and per APC for outpatient (2015-2024), covering about 3,000+ hospitals, released annually with roughly a one-year lag. Files reflect Original Medicare fee-for-service claims only - Medicare Advantage stays are excluded.