Medical coding
Dictionaries, how terms are matched, your decisions, licences.
Upload a listing of concomitant medications, medical history or adverse events exported from any EDC (CSV, TXT or XLSX), choose a dictionary and get every distinct wording coded or put in front of you for a decision. No AI is used: the rules below are the whole method.
How a term is coded
- Your own earlier decision for the same wording (case, punctuation and the ® sign are ignored).
- An exact name in the dictionary: the term itself, an inclusion term, an INN or a trade name.
- For drugs, the same after dropping dose, units and dosage form: “Аторис 20мг таб.” becomes “аторис”.
When a name leads to several codes (aspirin: B01AC06 or N02BA01, methotrexate: L01BA01 or L04AX03) the term is marked Several codes: pick by the indication. When it matches a group that may not be assigned (ICD-10-CM E11) you see the specific codes, the “unspecified / without complications” one first. Close spellings and terms with the same words are only suggested; nothing is applied without you.
Your decisions
Pick an option or type a code and press “Save my decisions”. The decision goes to your synonym list for that dictionary and is applied automatically next time. Every decision is written to the audit trail; you can remove one on the dictionary page.
Coding in a DM project
Inside a DM project open Medical coding in the project menu. Listings coded there are kept in the project: only the distinct wordings of the chosen column and their counts, never the file. Decisions belong to the whole team, are reused on every listing of the project coded with the same dictionary, and each one is written to the project audit trail. Data managers and the project manager code and decide; viewers read. The menu shows how many terms of the latest listing still wait for a decision.
A dictionary you uploaded under your licence stays yours inside a project too: colleagues see the figures of listings coded with it, never its terms or codes. Built-in open dictionaries are shared by everyone.
Dictionaries
- ICD-10-CM 2026 (USA, public domain): built in, about 47 000 codes. Only the most specific codes can be assigned; seventh-character extensions are not expanded yet.
- CTCAE 5.0 (NCI, public domain): built in, 837 adverse event terms with grades 1–5 and definitions, plus our own synonyms for British spellings and abbreviations (Anaemia, ALT increased). Term codes are the MedDRA codes NCI publishes in CTCAE; MedDRA itself is not included.
- UCUM units: built in, common clinical units with Russian and English spellings (ммоль/л, Г/л, 10^9/л, мм рт. ст.) and all units of the UCUM essence. Unit case matters: Г/л is 10*9/L, г/л is g/L; when case differs only, the match is marked.
- ICD-11 and LOINC are free, but their licences allow sharing only unchanged copies with attribution, and their sites need a download: load the WHO SimpleTabulation zip or the LOINC table (loinc.org, free account) in “Dictionaries”.
- ATC (WHO), demo extract: 50 common substances for trying the tool, with Russian and trade names. The WHO terms allow non-commercial use with a reference to the WHO Collaborating Centre for Drug Statistics Methodology; for real studies load the full index or a ГРЛС export.
- Loaded by you: the full ATC index, drug names → ATC from the State Register of Medicines (ГРЛС), ICD-10 from the Ministry of Health reference service (НСИ) or any code list, and MedDRA (ASCII zip from MSSO).