Terminology is not classification
Detailed clinical recording and statistical aggregation make different trade-offs. Mapping from one to the other is a purposeful transformation and may discard distinctions by design.
26terminologies and semantic patterns
12semantic use contexts
12problem-led scenarios
16evidence-led findings
1. Executive judgement
The difficult semantic decision is rarely “which code system?” It is deciding what kind of meaning must be preserved.
Clinical terminologies, classifications and product dictionaries are designed for different jobs. A terminology can preserve rich clinical distinctions, a classification can intentionally collapse them for comparison, and a value set can constrain what is appropriate in one context. Mature semantic architecture lets those artefacts coexist while governing every mapping between them.
A code system is not a value set
Code systems define concepts. Value sets select which codes may be used in a specific context. Treating them as interchangeable creates drift.
Version is part of meaning
Longitudinal records must retain which release, subset and mapping were in force when data was created or transformed.
2. Method and evidence
Semantics is treated as operational infrastructure, not a vocabulary catalogue.
The supplied Health Data Operability material provides the core distinction among terminologies, classifications, semantic debt, metadata, preservation and multilingual governance. Current authoritative publisher sources verify fast-moving release information.
3. Semantic stack
Meaning passes through several governed layers.
Concepts & definitions
Stable identity and definition independent of applications.
Code systems & classifications
Terminologies, dictionaries and classifications provide identifiers and structures for different purposes.
Value sets & bindings
Use cases narrow broad terminologies into canonical, versioned subsets and bind them to fields, profiles and forms.
Mappings & mediation
Crosswalks relate legitimate representations while making direction, equivalence and loss explicit.
Services, versions & preservation
Terminology services operationalise semantic rules; historical releases and mappings keep longitudinal meaning recoverable.
4. Semantic landscape
Clinical fidelity and statistical aggregation are different axes of value.
This schematic is an analytical navigation aid, not an authoritative taxonomy. Select a node to open its catalogue entry.
5. Interactive semantic catalogue
Select terminology by semantic purpose.
Suitability changes with use context. Current use and strategic potential are qualitative professional assessments rather than measured adoption statistics.
6. Decision aid
Build a semantic stack around the use case.
Six questions before coding
7. Semantic operations
A terminology service turns governance into executable infrastructure.
LookupResolve a concept, display, definition, properties and status.
ExpandGenerate current membership of a computably defined value set.
ValidateCheck whether a code is valid for a value set, version or historical date.
TranslateApply a governed directional mapping.
SubsumptionTest ancestor/descendant relationships where the terminology supports it.
DesignationResolve preferred labels and synonyms by language or usage.
Version resolveUse the release that applied when historical data was recorded.
EvidenceRecord which terminology, value set, mapping and service version produced the decision.
8. Governance and stakeholder analysis
Semantic governance is federated expertise with coordinated publication.
Domain stewardship
Clinical and professional owners remain accountable for meaning.
Shared terminology service
Publish authoritative releases, value sets and mappings once so systems do not maintain divergent copies.
Multilingual governance
Translation and language designations need clinical semantic review, not just UI localisation.
9. Semantic anti-patterns
Semantic debt is usually created by decisions that look locally reasonable.
The message validates
Conformance equals meaning
Structural validation is mistaken for semantic equivalence.
Close enough
Approximate mapping
A convenient target code is chosen without residual mismatch or confidence.
Everyone uses SNOMED
Ungoverned vocabulary choice
An entire terminology is allowed where a governed value set was required.
One code column
Systemless coding
Codes are stored without canonical system identity or version.
Make history current
Retrospective recoding
Old codes are silently replaced with modern ones.
Translate the label
Linguistic flattening
Literal text translation is assumed to preserve clinical meaning.
One canonical vocabulary
Semantic monoculture
Specialist or regulatory semantics are flattened beyond the chosen vocabulary's scope.
Mapping lives in ETL
Hidden mediation
Semantic knowledge survives only as unversioned code or supplier configuration.
Classification at source
Premature information loss
Coarse reporting categories replace richer clinical capture.
10. Interactive findings register
Key findings and evidence type.
| ID | Dimension | Finding | Evidence type | Severity | Confidence | Sources |
|---|
11. Recommended actions
Treat semantics as shared infrastructure.
Separate code systems from value sets
Give each governed subset a canonical identity, owner, version and change process.
Version mappings
Record direction, purpose, source and target versions, equivalence and review evidence.
Operate a terminology service
Support lookup, expansion, validation, translation and historical resolution as reusable services.
Preserve source semantics
Keep original codes and versions when deriving classifications or analytical representations.
Link metadata and terminology
Connect data definitions to code systems, value sets, units and model bindings.
Measure semantic debt
Track local codes, approximate mappings, divergent value sets and unversioned transformations.
12. Source register
Sources and how they were used.
13. Limitations
What this navigator does not claim.
Terminology adoption varies by jurisdiction and speciality. The catalogue is selective, not exhaustive. Licensing and national implementation requirements must be checked before deployment. Exact code-system releases, value sets and mappings should always be resolved from the authoritative publisher or terminology service at implementation time.