Institutions and data teams
Turn your research into a foundation for progress.
- Publish under your own name
- Set access and review requirements
- Keep ownership and get credit
AI for real-world impact
Accelerating AI for human progress.
SPHERE brings researchers, institutions and AI builders together to advance AI in fields that improve people’s lives—from healthcare to scientific discovery. Find the data, build on research and test ideas against real-world evidence.
Progress needs more than better models. It needs shared knowledge and real-world evidence.
Important questions in healthcare, science and other fields depend on data and expertise spread across institutions. Researchers need ways to work together without giving up control of sensitive information.
SPHERE connects that work: discover datasets, organize studies and evaluate ideas with the people who know the data. The goal is to shorten the path from AI research to useful applications.
From data to discovery. With owners in control.
- 01
Keep original data under your control
Prepare research data in your own environment and choose what can be shared.
- 02
Publish a protected research copy
Check a SPHERE copy and publish a version with its documentation, access terms and supporting research.
- 03
Build, test and validate
Explore published copies and develop ideas. Request owner-approved runs on original data when real-world validation is needed.
Everyone brings a piece.
Researchers
Build on the evidence your question needs.
- Search by question, not keyword
- Preview data and review its access terms
- Publish your evaluations, papers and results back
AI builders
Test your model beyond the data you already have.
- Find datasets relevant to your task
- Request validation on original data
- Compare results across studies and groups
Growing on SPHERE
See everythingHospital Quality Atlas 2025Every Medicare-certified U.S. hospital, by name and location: city, state, ZIP code and county, with its type, ownership, emergency services, birthing-friendly status and CMS overall star rating.County Chronic Disease Lens 2023Model-based prevalence of diabetes, obesity, high blood pressure and uninsured adults for every U.S. county, with 95% confidence intervals (CDC PLACES 2023).Four-Site Heart Disease CohortClinical measurements and angiography outcomes for 920 patients from Cleveland, Budapest, Zurich and Long Beach, with a site column for cross-hospital testing.Wisconsin Diagnostic Breast Cancer BenchmarkThirty image-derived nucleus features and the diagnosis for 569 fine-needle aspirates: a standard benchmark for diagnostic models.
Build AI that moves us forward.
Bring a question, a dataset or a model. Help advance AI in a field that matters to people.
Free for non-commercial academic research during the research preview. One account for the apps, portals and API.