Skip to main content
AI-assisted curation, review and search

The Library

Every AI use case in healthcare — documented, classified, comparable.

The AIH Lab Repository organises real-world healthcare AI into 18 clinical families, so any clinician, hospital, regulator, or researcher can find, compare, and learn from what works.

Published use cases
27
Validated + scaled
6
Countries represented
6

Healthcare AI arrives in fragments. AIH Lab gives it a map.

Three forces make healthcare AI hard to use today — and AIH Lab addresses each one structurally.

Fragmentation

Every hospital documents its AI differently. Comparing across organisations is nearly impossible.

Regulation

The EU AI Act made healthcare AI a high-risk category. Most hospitals lack the structure to show classification and oversight.

Trust

Clinicians, patients, and the public need clear answers about what AI is doing, with what evidence, under what oversight.

Read our approach

18 use case families

Healthcare AI does not arrive in neat packages. A family is a clinical concept; every local deployment is an instance of it. The taxonomy makes hundreds of disconnected records comparable.

Open the family map

From submission to validated deployment

Every use case travels a structured pipeline. Each stage builds trust, evidence, and replicability.

See each stage in detail →

  1. 1

    Submit

    Any stakeholder registers a use case instance.

  2. 2

    Curate

    The AIH Lab team reviews completeness and confirms the family.

  3. 3

    Mature

    In-depth clinical, technical, ethical and operational review.

  4. 4

    Validate

    Formal real-world validation and the Assurance Pack.

  5. 5

    Scale

    Packaged for replication across other organisations.

One platform, every stakeholder

What you see first depends on who you are. The data is the same; the AIH Lab surfaces the part of it you need.

AI assists the work — humans decide.

AIH Lab uses AI in three places: completeness checks at submission, family suggestions during curation, and synthesis at the Hub. Every machine output is labelled. Every decision is human.

How the AI works

Built for trust

Three structural commitments shape every feature.

  • No patient data — deployment metadata only.
  • Full audit trail — every transition logged.
  • Aligned with EU AI Act — Assurance Packs map to obligations.
Our governance approach

Evidence status is always explicit

No record appears without a trust badge derived from its pipeline stage. You always know what you are looking at.