Please join us on July 25th noon EST for Vocabulary WG call happening in CDM Teams channel (link here) where will discuss deduplication and harmonization efforts around LOINC/SNOMED measurements with our amazing guest speaker @Polina_Talapova!
Was great to see almost 30 people and great discussion about measurement hierarchies and measurement-unit pairs. If you missing the call, slides and recording are available in the Files in Vocabulary subgroup Teams channel.
Bottom line: Polina and others worked hard to link some of the LOINC and SNOMED measurements in one hierarchy so that you can select a SNOMED code + all descendants to get a nice measurement concept set, but you still always need to look at the concept set closely since the hierarchy is not comprehensive. A great opportunity to collaborate to create more links (as @MaximMoinat pointed the gap with SNOMED red blood cell count and LOINC erythrocytes count)! If you have good measurement concept sets where you caught gaps in hierarchies - send it our way
Another bottom line: everybody knows measurement-unit pairs are a pain, in ETL and in cohort construction. It would be awesome to know which units go with which measurements potentially leveraging amazing work by @Vojtech_Huser here and N3C/@stephanieshong work here. Thanks @DaveraG for a great question and hoping to talk more about in Themis (@MPhilofsky).
Yes, Themis would like to help co-ordinate and bring about some conventions for these data. @Dymshyts and the DQD group (@katy-sadowski @clairblacketer) have already incorporated some measurement & unit pair checks into the DQD.
I think the best first step would be to gather all on a call to discuss what has been done and what the goals are for these data. Themis meets the 1st and 3rd Thursday of the month at 9:30am Eastern Time. Does Thursday, September 7th work for those interested?
This looks pretty useful, we can reuse some results of this work in the DQD right away:
canonical_units_of_measure can be used to check plausible values. It has plausible ranges for 53 measurements and their canonical units, and we can add more units using it along with a dictionary of conversion (TableS3.csv)
i’m out of town on 9/7 but could catch up on the recording so as not to hold up the discussion (as long as Dima and/or Clair could attend )