Ariadne 
Ariadne is a Python toolkit to harmonize source vocabularies into the OHDSI Standardized Vocabularies.
It currently supports three workflows:
- Conditions: exact matching and hierarchy matching (for identifying a parent concept when no exact standard match is available).
- Drugs: prepares source codes for inserting into the OHDSI Boiler software as described here.
- Procedures: exact source-to-standard concept mapping for procedure vocabularies.
The toolkit is designed to be extensible, allowing for the addition of new workflows (e.g., measurements, devices) and mapping strategies as needed.
Features
- Clean-up: normalizes source terms per mapping rules, removing non-essential information.
- Verbatim term mapping: maps terms that (almost) exactly match standard concepts using normalization (lowercasing, punctuation removal, stemming).
- Embedding vector search: retrieves semantically similar standard concepts as mapping candidates.
- LLM-assisted exact mapping: selects the best standard concept from candidates.
- Hierarchy matching (Conditions): finds parent concepts for unmatched terms in the standard hierarchy.
- Evaluation: compares outputs against gold-standard mappings.