Skip to inventory
CREED.AIPrimum non nocere

A CREED.AI flagship public initiative

Health.AI.DepO

A CREED.AI Deployment Observatory

A living public inventory of how artificial intelligence is being used in health—across countries, health systems, communities and stages of evidence.

Explore deployments
Health.AI.DepOA CREED.AI INITIATIVE
01Tried
02Tested
03Effective
04Beneficial
TRIEDTESTEDEFFECTIVEBENEFICIAL

From innovation to health benefit

Make deployment
visible.

The world cannot learn from health AI projects it cannot see.

Health.AI.DepO documents the product, place, population, purpose, evidence and implementation experience behind each real-world deployment. It makes progress visible, distinguishes claims from demonstrated outcomes, and reveals where proven benefits are still failing to reach people.

Public deployment inventory

Explore the field.

4records

2countries

4health areas

GEOGRAPHIC VIEW4 matching deployments

Geographic markers show the approximate location of projects, not patient-level data.

DEPLOYMENTS4 results

A transparent evidence pathway

No single
“proven” label.

Evidence is specific to a use, population, workflow and setting. Health.AI.DepO shows what is known, how it is known, and what remains uncertain.

  1. 01

    Tried

    Used or evaluated in a defined real-world setting.

  2. 02

    Tested

    Assessed with a protocol, comparator, or structured outcome measures.

  3. 03

    Effective

    Evidence shows a meaningful result for a specified use, population and context.

  4. 04

    Beneficial

    Benefits are sustained in practice, with safety, equity and implementation monitored.

Every claim needs a source.

Vendor-reported, implementer-reported, peer-reviewed and independently evaluated evidence are identified separately. Unreviewed submissions remain in a private moderation queue.

Build the shared picture

Know of a health AI
deployment?

Researchers, health systems, governments, community organizations, developers and funders are invited to contribute. Successful, inconclusive, paused and discontinued projects all help the field learn.

Do not submit patient-level or personally identifiable health information.