Models
Healthcare-specific and healthcare-adapted AI models.
HAIHQ / Open research
Open research and infrastructure for Health AI.
HAIHQ is a nonprofit, open-source research organization building models, datasets, benchmarks, and tooling for Health AI.
We research technical problems, build systems to investigate them, evaluate what works, and make the resulting work open.
What We Work On
Datasets
Open healthcare datasets for training, testing, and studying Health AI systems.
Benchmarks
Evaluation systems for measuring capability, reliability, safety, robustness and real-world behavior.
Tooling
Open infrastructure for building, testing and operating Health AI systems.
Current work
ReFHIR
Reactive FHIR infrastructure for applications and AI agents.
ReFHIR is an open-source FHIR server whose query results stay live. Applications and AI agents ask once, and the current answer arrives again whenever the underlying data changes.
Instead of manually combining queries, subscriptions, refetching and state reconciliation, the server keeps each result current. It is a full FHIR server underneath, released under Apache-2.0.
Research through building.
- Research
We start with a technical question.
- Build
We build the system needed to investigate it.
- Evaluate
We measure what happens.
- Open
Then we release the useful parts openly.
Get Involved
HAIHQ works with researchers, engineers, clinicians, healthcare organizations, universities and supporters interested in advancing open Health AI.
Contribute · Collaborate · Sponsor
Get involved →