search

TranslatorSRI / Babel

Babel creates cliques of equivalent identifiers across many biomedical vocabularies.
MIT License
9 stars 2 forks source link

ICD-11 Mappings #357

Open beasleyjonm opened 1 month ago

beasleyjonm commented 1 month ago

I'm encountering ICD-11 disease identifiers which are not normalizable currently. Is there any ongoing effort to create these mappings in Babel? I noticed that there is a makefile for MONDO ingest that handles ICD-11 codes. (https://github.com/monarch-initiative/mondo-ingest/blob/main/src/ontology/mondo-ingest.Makefile)

Could we update diseaseorphenotypicfeature compendia to include ICD11 prefix codes?

gaurav commented 1 month ago

I think we can. But there are a few problems (but also a few solutions!):

  1. Biolink Model doesn't support ICD11 as a prefix for Disease. I've opened a ticket with them to make this change: https://github.com/biolink/biolink-model/issues/1531
  2. It looks like MONDO does have some ICD11 mappings, although they're a little weird: for example, MONDO:0004979 "asthma" has an xref to http://id.who.int/icd/entity/1656445230, which corresponds to code CA23 Asthma. Would http://id.who.int/icd/entity/1656445230 (or icd11.foundation:1656445230) be an acceptable identifier for you, or are you looking for ICD11:CA23? It looks like MONDO has a significant number of the former, so this might be all we need.
  3. There are some ICD-10 to ICD-11 mappings available for download, so we might also be able to sidestep MONDO if we need to and just map from ICD-10 directly to ICD-11. This also includes codes like CA23 for asthma, so this could be a way for us to map to this information as well.

How quickly do you need this done?

beasleyjonm commented 1 month ago

I think this is relatively high priority for the MATRIX Everycure project. Particularly because they are trying to get target->disease edges from Therapeutic Targets Database (https://idrblab.net/ttd/), which uses ICD11s for the disease terms. We could probably do disease names->Name Resolver-> normalized IDs, but I think the better strategy is to try to get ICD11<->ICD10 mappings in Babel.

So I like the strategy of just using the ICD-10 to ICD-11 mapping files while also updating the Biolink model as suggested in your ticket. Thanks for looking into this!