Standards & Interoperability
SNOMED CT (Systematized Nomenclature of Medicine — Clinical Terms)
SNOMED CT is the most comprehensive, multilingual clinical terminology in the world. It assigns a unique, machine-readable code to hundreds of thousands of clinical concepts — diseases, findings, procedures, body structures, organisms, and substances — and links them together in a rich web of relationships. Mandated for use across the NHS and adopted in over 50 countries, SNOMED CT is the vocabulary that makes semantic interoperability possible: it lets two systems, and increasingly an AI model, agree on exactly what a clinician meant.
What SNOMED CT actually is
At its core, SNOMED CT is a graph of concepts. Each concept has a unique numeric identifier (for example, 73211009 means 'Diabetes mellitus'), one or more human-readable descriptions, and defined relationships to other concepts — most importantly the 'is-a' hierarchy that places diabetes under 'disorder of glucose metabolism' and so on up the tree. This structure means software can reason about clinical data: a query for 'diabetes' can automatically include every subtype because the hierarchy encodes that they are kinds of diabetes. That is something a flat list of codes can never do.
SNOMED CT vs ICD-10
The two are complementary, not competing. SNOMED CT is a clinical terminology — designed to capture exactly what a clinician observed and did, at the point of care, in fine detail. ICD-10 is a classification — designed to group conditions into broader categories for reporting, statistics, and billing. A patient's record might be documented in SNOMED CT and then mapped to ICD-10 for a claim or a public-health return. A well-built health data platform usually stores the granular SNOMED CT detail and derives ICD-10 (and other classifications) from it.
Why SNOMED CT matters for AI
AI features in healthcare live or die on structured meaning. When a clinical NLP pipeline extracts 'patient reports crushing chest pain' from a note, mapping that phrase to the correct SNOMED CT concept is what turns free text into something a triage agent, decision-support rule, or analytics dashboard can act on reliably. SNOMED CT also gives large language models a controlled vocabulary to ground their outputs against, reducing ambiguity and enabling auditable, code-backed reasoning rather than free-text guesses.
Working with SNOMED CT in practice
Using SNOMED CT well means handling its scale and its subsets. The full edition is enormous, so implementations rely on reference sets (subsets curated for a use case or specialty) and value sets to keep things manageable. Mapping local or legacy codes to SNOMED CT is a recurring data-engineering task, and keeping up with the twice-yearly releases matters for accuracy. Licensing is free in member territories (including the UK via NHS England) but should be confirmed per deployment.
Frequently asked questions
Is SNOMED CT free to use?
In SNOMED International member countries — including the UK, US, and Australia — SNOMED CT is available at no additional licence cost for use within that territory. Always confirm the specific licensing terms for your jurisdiction and use case before deployment.
What's the difference between SNOMED CT and LOINC?
SNOMED CT covers clinical concepts broadly (diagnoses, findings, procedures), while LOINC specialises in identifying laboratory tests and observations. They are frequently used together, with LOINC naming the test and SNOMED CT coding the clinical result or finding.
Why not just use free text instead of SNOMED CT?
Free text is easy to write but hard for software to use reliably — it can't be searched, aggregated, or reasoned over consistently. SNOMED CT codes give the same clinical meaning a stable, machine-readable identity, which is essential for analytics, decision support, and trustworthy AI.
Building features that depend on clean, coded clinical data? We design SNOMED CT mapping and terminology pipelines. Book a discovery call to scope it.