OHDSI Home | Forums | Wiki | Github

How to map local specific terminology into standardized concepts?

Hi All,
I am new to the OMOP CDM and its standardized vocabulary. My background is in library science and I am working with computer programmers to develop our databases for academic medical research. I have a local specific terminology called “Ontario Health Insurance Number”. It is an essential field of the patient demographic table in my database. Am I correct in thinking that I should use the ATLAS to create a new concept set with two standardized SNOMED Concept IDs: “Ontario” + “Health Insurance Number”?
Many thanks for your help in advance!

Welcome to the OHDSI community, Cordelia He!
Perhaps you should start your OHDSI journey with this book. You can read about OHDSI’s open-source tools here (The Book of OHDSI). This book is intended for both OHDSI newcomers and veterans alike, and aims to be practical, providing the necessary theory and subsequent instructions on how to do things.
In this book you’ll read about mapping - the process to transform one source concept into a Standard one. All codes used in the source databases have to be translated to Standard Concepts.
For mapping we use such tool as Athena (Athena). You can see how Athena works in this video (10-Minute Tutorials: ATHENA (Michael Kallfelz, April 13 OHDSI Community Call) - YouTube).
Enjoy your journey and feel free to ask questions on the forum!

Hi Varvara,

I have completed reading from Chapters 1-7 of The Book of OHDSI, watched the tutorials of ATHENA, ATLAS, 2016, 2019 and 2020 Symposium tutorials. At the moment, I have a basic concept about how to map a non-standardized source concept into a Standardized Concept.
Now I am in my “hands-on experience” stage. My challenge is: On my “Patient Demographics” table, there is a field called “Last Name”. On ATHENA, here are my search outcomes:


I only found a “Standard Concept ID” from LOINC, but not SNOMED. Should I go ahead to use this Concept ID of 3046810 even though it is from LOINC?

My biggest challenge is how to handle all the locally specific terminology that is essential to my Care Site only (e.g. Ontario Health Insurance Number, and many medical terminologies that even the clinicians disagree with each other). I have learned how to create a complex concept set with the help of OMOP PHOEBE. My next step is to learn the ATLAS “New Cohort”.

Since I am the only one in my group who is currently studying the OMOP CDM model in full detail and I am only giving less than 2 weeks to have a solid progress update, it will be great if I can learn from other contributors who are also adopting the OMOP tools to map their concepts. Please forgive me if I ask some simple and stupid questions.

Thank you very much!

After watching the OHDSI’s 2019 Tutorials - OMOP Common Data Model and Standardized Vocabularies (Full Tutorial) on YouTube for the 3rd time, I guess I have finally got the idea about how to make this work :sweat_smile:

hi Cordelia,

Kindly share the link (OHDSI 2019 tutortial) and if possible how you plan to deal with local specific terms into standard concepts. I have a similar situation where in we use e.g. ‘Other Race’ and there is no standard concept for that in OMOP.

Looking forward to hearing from you.

Thanks
Sanjna

Hey Sanjna,

Here is the link: https://youtu.be/vHMkBaHJrDA
It is under the OHDSI YouTube Channel.
I have been watching and learning from almost all the video tutorials alongside with the eBook of OHDSI.
You might also want to watch the 2020 Symposium tutorial on creating concept sets with the help of OHDSI PHOEBE: https://youtu.be/1YcZKOldMrI (10:59 - 32:07 - Anna Ostropolets (Concept Prevalence — an OHDSI Network Study; Design Diagnostics — PHOEBE)

If my understanding is correct, a non-standard source concept can be mapped to SNOMED like this. I use “lesbian” as an example:

So the OMOP “Concept_ID” for my source concept “lesbian” should be: “4230492”. I am not a programmer, so I primarily rely on the ATHENA website to look up Concept_ID.

I hope this helps.

Cheers,
Cordelia

Thanks for the reply.
However, my scenario is a bit different. There is no concept id (either standard or non-standard) for my Race Value i.e. ‘Other Race’. But thanks for sharing the information, appreciate your help. :slight_smile:

Regards,
Sanjna

ATLAS is a good tool for Cohort building. I am not sure whether we need to install the atlas files @git to get started or we need all 3 i.e. Web API + Achilles + Node.js to run atlas. Let me know once u get hang of it. Thanks!

-sanjna

There isn’t, @Sanjna. “Other” is a so-called flavors of null. Meaning, you don’t really know what race it is. “Other” is bad in particular because it assumes other than some defined list. What is that list?

Bottom line: An analytics application cannot use “other” in any meaningful way, ie.it has no meaning for our purposes. We therefore don’t map it to anything with meaning. Write 0 into the concept_id field.

Can you start another post, @Sanjna? In “Implementers”?

Thank you so much @Christian_Reich . Makes sense.

t