Welcome to AI & SNOMED CT
This is the SNOMED community’s space for talking about artificial intelligence and SNOMED CT: what works, what doesn’t, and what needs a pinch of salt.
What belongs here
- Applications: LLMs, embeddings, NLP, coding and mapping assistants, semantic search, clinical text processing, and anything else that uses SNOMED CT as an input, output or grounding source.
- Tools and projects: open-source libraries, MCP servers, plugins, datasets and prototypes you’ve built or found useful.
- Evaluation and benchmarking: methods, results, failure modes, and how you measure “correct” when a model confidently invents a concept ID.
- Terminology for AI: how SNOMED CT content, modelling and ECL can improve AI outputs, and where the terminology needs to evolve.
- Safety, governance and ethics: clinical risk, bias, transparency, and regulatory considerations.
- Questions and war stories: including the “we tried it and it went badly” posts. Those are often the most valuable.
What doesn’t belong here
- General AI news with no SNOMED CT angle. There are plenty of places for that.
- Product advertising. Sharing something you’ve built is welcome. A sales pitch is not.
- Questions about SNOMED CT content, modelling or releases. Those belong in the relevant categories elsewhere on the forum.
- Anything that needs a support ticket. Use the usual channels for that.
Ground rules
- Show your working. If you report results, tell us the model, version, prompt approach, SNOMED CT edition and release date, and how you evaluated. “It works great” is not a finding.
- Say when AI helped you write it. Use AI to draft if you like, but check it, and tell us when it was involved. Unverified generated text, particularly invented concepts or codes, will be removed.
- Be constructive. Challenge ideas, not people. Scepticism is welcome, and so is enthusiasm.
A note on status
Posts here are community discussion. Unless a post is clearly marked as coming from SNOMED International, nothing here is an official position, endorsement, or guidance. Where SNOMED International staff share official updates, we’ll say so.
Getting started
Introduce yourself in the replies below: who you are, what you’re working on, and what you’d most like to learn. When you start a new topic, please tag it (for example llm, nlp, mapping, evaluation, tooling, safety) so others can find it.
Right, over to you.