Instant access to anonymized healthcare data.
Search EHR, claims, surveys and environmental data in one catalog. See coverage, tables and variables before you request access.
100+
Datasets
20B+
Records
1980–2026
Coverage
Population Health
2011–2025
Behavioral Risk Factor Surveillance System (BRFSS)
State-level prevalence of chronic-disease risk factors from the CDC’s yearly phone survey.
3.2M records
·
State-level
Geography & Environment
1980–2020
EPA Air Quality
Daily outdoor pollutant readings from 4,000+ US monitors: ozone, PM2.5, NO2, SO2, CO and lead.
41.8M records
·
Monitor-level, daily
Join 2,000+ researchers exploring Healthcare Datahub
Less time waiting for data. More time doing research.
Go from a research question to the right dataset in minutes. Check fit before you request anything, then work with the data in a secure workspace.
Find the best dataset for your study
Search every dataset by condition, variable or code, and compare population, coverage, granularity and tables side by side.
Coverage years and row counts
Coding systems and variables
Plain-language summaries
Browse the data catalog
Data catalog
Example search
Surgical patients with vitals and labs
INSPIRE Dataset for Perioperative Medicine
Strong match
Patient-level · 2011–2020 · 144M records
Diabetes 130-US Hospitals
Labs, no vitals
Encounter-level · 1999–2008 · 101.8K records
PMC Patient Notes
Narrative text
Patient narratives · 167K records
Mira AI
Connected to the catalog
Which dataset has HbA1c results for hospitalized patients with diabetes?
Diabetes 130-US Hospitals covers 101,766 inpatient encounters from 1999 to 2008. HbA1c results are in the A1Cresult column, alongside readmission flags and diabetes medication changes.
Diabetes 130-US Hospitals
Ask Mira AI anything about the data
Mira AI is our dataset agent. Ask in plain language and it answers with the exact dataset, table and variables that fit, so you know in minutes whether we have your data.
Cohort feasibility
Variable lookup
Dataset comparisons
Chat with Mira AI
Request access. Analyze in a secure workspace.
Tell us about your project in a short form. We reply within two business days with access options, then you work in a cloud workspace without moving patient-level data.
Python, R and SQL
Data stays in the Hub
Reply within two business days
Request access
Your project
Secure workspace
Request
Short form
Approved
Access options
Analyze
Python · R · SQL
-- INSPIRE: median length of stay by ASA class
SELECT asa, APPROX_QUANTILES(los, 2)[OFFSET(1)]
FROM inspire.operations
GROUP BY asa ORDER BY asa;
Explore datasets by data type
Pick a data type to see what’s inside, what it’s good for, and the datasets you can request today.
Clinical & EHR
18
Claims & Utilization
14
Population Health
19
Providers & Quality
18
Geography & Environment
12
Clinical & EHR
18
Datasets
11.4B+
Records
1999–2026
Years
What’s inside
EHR encounters · Labs · Medications · Imaging & reports · Waveforms · Genomics
Good for
Outcomes research · Prediction models · Phenotyping · Clinical NLP · Imaging AI
Diabetes, Inpatient, Readmissions
1999–2008
Diabetes 130-US Hospitals (1999–2008)
101,766 anonymized inpatient encounters for patients with diabetes across 130 US hospitals.
101.8K records
·
Encounter-level (anonymized)
Surgery, Anesthesia, Critical care
2011–2020
INSPIRE Dataset for Perioperative Medicine
About 130,000 anonymized surgical cases with OR, ward and ICU vitals, labs, medications and outcomes.
144M records
·
Patient-level (anonymized)
Clinical NLP, Case reports
Not time-bound
PMC Patient Notes
167,000 free-text patient summaries from published case reports, linked to articles and similar patients.
167K records
·
Patient narratives
Your dataset agent
Chat with Mira AI. Get specific answers about your data.
Mira AI knows every dataset in the Hub: its tables, variables, coding systems and coverage. Ask it anything and it answers with the exact dataset and fields, before you request access.
Is my cohort here?
Describe your study and Mira AI tells you which datasets cover it.
Which variables exist?
Find the table, column and coding system behind each concept.
How do datasets connect?
See which datasets join on FIPS, ZIP, NPI or NDC codes.
Mira AI
Connected to 18 datasets
Which datasets can I join at the county level?
Three datasets join on county FIPS codes: Medicare Market Saturation & Utilization (2014–2025), Social Deprivation & Vulnerability Indices, and USDA Rural–Urban Codes. EPA air-quality monitors also carry state and county codes.
Try asking
Do you have perioperative vitals linked to procedure codes?
Is there overdose mortality data after 2022?
Ask about datasets, variables or feasibility…
Anonymized data, secure by design
Every dataset in the Hub is anonymized or aggregated before it reaches you, and you analyze it inside a secure workspace.
Anonymized records
Patient-level data is de-identified before it reaches the Hub, and dates can be shifted to protect privacy.
Small counts hidden
Dataset profiles never show counts under 11.
Secure workspace
Analyze approved data in the Hub’s cloud workspace without moving patient-level records.
Reviewed access
Every access request is reviewed before a workspace opens.
Frequently asked questions
Answers about the data, Mira AI and how access works.
Still have questions?
Ask Mira AI about the data, or email hi@healthcaredatahub.ai.
The healthcare data you need, without the months of waiting.
Explore anonymized EHR, claims, population and provider datasets, and see what your next study could look like.