Map Drugs: EMA Products

Map EMA centrally authorised product presentations to RxNorm standard concepts.

When to Use

When asked to map an EMA product to RxNorm, or create/review a mapping.tsv in an EMA product folder.

Finding Unmapped Products

To list all product folders that have parsed data but no mapping.tsv:

bash

python3 skills/map-ema-drugs/find_unmapped.py

This prints folder names (one per line) that need mapping. Use it to find the next product to work on.

Prerequisites

Refer to the find-concepts skill for how to search for RxNorm concepts, and the map-drugs skill for general mapping principles.

Input Files

Each product folder under data/ema/products/<ema_number>/ contains:

•ema-info.txt: Product metadata (medicine name, active substance, INN, therapeutic area)
•parsed_data*.tsv: One row per authorised presentation with ma_number, strength, pharmaceutical_form, route_of_administration, packaging, content, pack_size, product_name, active_substance, etc.

Read both files to understand the product before mapping.

Output Format

Create mapping.tsv in the product folder following the schema in the map-drugs skill. The source ID column is ma_number (EU marketing authorisation number from parsed_data.tsv).

One row per ma_number in parsed_data*.tsv. Every presentation must be mapped.

If a mapping.tsv already exists, check the mappings and note any needed changes in flag.tsv.

Biosimilars

If ema-info.txt contains Biosimilar: Yes, follow the biosimilar mapping rules in the map-drugs skill.

Validation

After creating or editing mapping.tsv, validate it:

bash

python3 skills/map-drugs/validate_mapping.py data/ema/products/<ema_number>/mapping.tsv

Fix any errors, then regenerate combined files:

bash

python3 skills/resolve-conflicts/scripts/regenerate_combined.py

Flagging Problems

If no suitable RxNorm results exist or mapping needs validation, add a flag.txt in the product folder explaining the issue.

EU-only Products

Some EMA products have no FDA equivalent and no RxNorm concept. Leave concept fields empty, populate suggestion with the ideal concept name, and create flag.txt.