Target Concepts

Why this step?

In the Mapping Editor, the right pane gives you two ways to find a target concept:

1. Browse a concept set from a reference dictionary (fast, focused).
2. Free search: lets you search across all OHDSI concepts, but picking the right one is harder given how many exist.

This tab populates both modes.

Concept Sets

What is a concept set?

A concept set groups several standard concepts under a single clinical label. For example: “Total bilirubin”, “Sepsis”, or “Norepinephrine”. In practice, each one points to the matching LOINC, SNOMED or RxNorm concepts — often numerous and technical.

Items and resolution

A concept set is a list of items. Each item is a concept (with a conceptId, a conceptName, a vocabularyId, a conceptCode, a domainId, a conceptClassId and a standardConcept flag) along with three options:

• isExcluded — the item is excluded from the list (useful for “all X except Y”).
• includeDescendants — automatically includes more specific concepts (descendants in the concept hierarchy).
• includeMapped — includes concepts that point to this item via concept_relationship.

A concept set can therefore be seen as an equation: a handful of base items, modulated by these options. Resolving a concept set means applying these options to obtain the final list of all retained concepts — a single item with includeDescendants can expand into several hundred concepts. This resolved list is what the Mapping Editor uses to propose targets.

Going further

• INDICATE — What are concept sets? — a hands-on introduction to concept sets, in the context of the INDICATE dictionary.
• OHDSI — Concept Set Specification — the formal specification of the JSON format used by OHDSI (and Linkr).
• ATLAS — Concept Sets wiki — the reference documentation for creating and managing concept sets in ATLAS.

Any concept set following the OHDSI format can be imported into Linkr — whether it comes from a thematic dictionary like the INDICATE Data Dictionary or is hand-crafted in ATLAS (OHDSI’s reference tool for building concept sets).

Importing a concept set

Click Import a Concept Set. Three sources are offered:

• Reference — bulk import from a referenced dictionary, for example the INDICATE Data Dictionary: all of the dictionary’s concept sets are imported in one go.
• URL — a direct link to the JSON file of a single concept set (typically a raw.githubusercontent.com/.../concept_sets/X.json file). Linkr keeps the link, so you can later update the concept set in one click if the source changes.
• File — a .json in OHDSI format ({ expression: { items: [...] } }).

Inspecting a concept set

The ⓘ icon at the end of a row opens a side panel with four tabs:

• Description — the long-form description written by the concept set’s authors. When the concept set comes from a dictionary like INDICATE, you’ll typically find the clinical definition, the coding choices made by experts, and guidance for concept mapping.
• Statistics — record and patient counts, distribution charts.
• Resolved concepts — the list of concepts retained once the concept set has been resolved (see above).
• Expression — the raw list of items in the concept set, with their conceptName, conceptId, vocabulary, domain, and three columns Ex / De / Ma (ticked when the item is marked isExcluded / includeDescendants / includeMapped).

Update, delete

• Targeted update — the refresh icon on a row re-fetches the JSON from the source URL.
• Update all — dedicated button, processes all concept sets that have a sourceUrl.
• Delete concept sets — click Edit: checkboxes appear, select the concept sets to delete then click Delete selected.

Filter the table

Concept sets are indexed by category, subcategory, provenance (source organisation) and version. Column headers offer dropdown filters. Handy when the catalog grows past several hundred entries.

OHDSI Vocabularies

This is where you load the full set of standard OMOP concepts, from an ATHENA download.

What to download?

On ATHENA, select the vocabularies you need (at minimum SNOMED, LOINC, RxNorm for standard usage) and trigger the bundle generation. Once it’s ready, you download it directly from the site as a ZIP file containing the OHDSI tables:

• concept — required, the full catalog of concepts used by the Mapping Editor’s search.
• The other tables shipped by ATHENA (concept_ancestor, concept_relationship, concept_synonym, concept_class, domain, relationship, vocabulary, drug_strength) are not used by the concept-mapping feature. You can keep them in the folder — they’ll be useful later for generating an OMOP ETL — but they aren’t needed here.

Load the folder

Unzip the archive, then in Linkr click Select folder and pick the ATHENA directory. Linkr identifies the known files automatically and indexes them in DuckDB.

Browse the vocabulary

Once the reference is loaded, you can browse the standard concepts directly from this sub-tab (search, filters by vocabulary / domain / class, “Standard only” toggle). But more importantly, the Mapping Editor unlocks its “Search” mode in the right pane — that’s where the reference truly pays off.