OHDSI Home | Forums | Wiki | Github

HSIG meeting: adding vocabularies to the CDM

General
#1

On Monday, July 31st, Anna Ostropolets will join the HSIG meeting to review the Vocabulary Team’s new process to add community sourced vocabularies to the CDM. She will explain which vocabularies can be added, where to add this info, when it will be incorporated, how it will be accessed, best practices, and other tidbits of helpful knowledge we should know.

The 2nd half of the meeting will be a discussion about creating a knowledge base for community members to contribute their mappings of source values/strings to standard concept_ids. Free text lab results, social/lifestyle facts, and flowsheet data are a few examples.

As per our usual meeting schedule, you can find us on MS Teams at 10am Eastern Time.

1 Like
#2

We are very interested on vocabularies around social care concepts such as when someone has received help in securing adequate food supply, sanitation, security, mobility, medication intake etc.

Missed the date, but may I ask what kind of thoughts are there at the moment on vocabularies and mappings around social/lifestyle facts?

#3

There is too much ‘opinion’ in mapping social constructs. What one organization thinks adequate food supply is, another will not. The opinion biases even shows up inside institutions.
This will be a headache to implement and even then, not give valid results.

#4

Thank you, a very good point. In addition to ‘opinions’ on measuring need, also the way to deliver social care differ from country to country. This can indeed be a problem.

We have used social service data in addition to health data in predicting individual, for example an elderly persons, future care needs. Adding up the individual results will tell the district for example how many intensive housing care units they will need to have available in the coming 12 / 36 months.

In Finland we have public health and social care and nationally agreed ways to measure who is in need for example of a home care service and who is not. So at least nationally there is not so much ‘opinion’ risk.

Would you have ideas on how we could map this social data into OMOP cdm? Is the only way creating our own vocabulary for it?

#5

There is but @MPhilofsky is much more qualified to answer the details than I.

1 Like
#6

@ElinaSarpola I replied to your direct message, but I will also reply here for others to see.

There are two ways to map these data into the OMOP CDM.

  1. Map the data to standard concept_ids using one of the methods listed in this poster.
  2. Create a custom vocabulary for these data and then insert the vocabulary and data into the CDM.

What you do really depends on your use case and the data. What questions do you want to ask the data? Are the standard concept_ids in the vocabulary a close enough match to the codes/terms in your data?

t