OHDSI Home | Forums | Wiki | Github

OHDSI vocabularies for patient record mining

I’m trying to setup the standardised vocabulary set in our text mining pipeline. I have created the set using OHDSI’s download process and created a script to flush it to our mysql server. In that process I found out that there is a table SOURCE_TO_CONCEPT_MAP that presumably contains the mapping to the original vocs. The data however is not contained in the download. It would be great to have it so that we can use this for our translation pipeline to other languages (e.g. Dutch). Where can I find this source mapping file?

@Erik:

It sounds like you are using vocabulary V5, correct? All the lexical information is in two tables only: CONCEPT and CONCEPT_SYNONYM. Everything in CONCEPT is English, and the Synonyms can be in any other language, but it’s mostly English as well. Could be non-US spelling variations. You can find the mapping between those Concepts representing source codes to those Concepts representing the Standard Codes in the table CONCEPT_RELATIONSHIP where relationship_id=‘Maps to’.

The table SOURCE_TO_CONCEPT_MAP is a remnant from V4, but is no longer needed in V5. However, it is still there as an empty shell for those folks who have their own local codes to translate.

However, there currently no Dutch

Yes, using V5. Ok, I didn’t check thoroughly on the documentation of the tables. I assume it is written somewhere. I want to use the original codes to link it to existing translations in other languages (e.g., Dutch).

What is an “original code”? There are three types of “codes” (we call them concepts): standard, classification and source. They are defined in the standard_concept field. Standard Concepts are the one mapped to from the source Concepts, and they are the only one used in the data. Which ones do you want to translate?

Let us know.

I want to translate (and include known translations ) for the standard concepts. I understand that all have been mapped from/to source concepts (which have the original source vocabulary assigned code).

Yes, That’s correct. Put the translations into the CONCEPT_SYNONYM field, with the language_concept_id=4182503 (Dutch language). Toss it over to us If you want to share with the community.

t