search

OHDSI / dbt-synthea

[Under development] A dbt ETL project to convert a Synthea synthetic data set into the OMOP CDM
https://ohdsi.github.io/dbt-synthea/
Apache License 2.0
16 stars 6 forks source link

feat: Improve `care_site_types` mapping #85

Open lawrenceadams opened 4 weeks ago

lawrenceadams commented 4 weeks ago

At the moment all care_site_types are hardcoded to 0. We can probably make this more realistic and map to proper

Perhaps we can use some simple heuristics like presence of PCP to bind the care_site_type to 38004247 | Ambulatory Primary Care Clinic / Center|.

I've not seen anyone use this attribute in real life in fairness (most of the work I've been a part of doesn't find it relevant)...

Somewhat related to #83

heschmidt04 commented 4 weeks ago

I like the idea of settling on a code. Question along those lines. Is our etl example supposed to represent ehr. If so, can we choose that?

On Wed, Oct 23, 2024, 17:34 Lawrence Adams @.***> wrote:

At the moment all care_site_types are hardcoded to 0 https://github.com/OHDSI/dbt-synthea/blob/47c7fd058680f38e2bb5858e28ec7c5f32b51c31/models/omop/care_site.sql#L4. We can probably make this more realistic and map to proper

Perhaps we can use some simple heuristics like presence of PCP to bind the care_site_type to 38004247 | Ambulatory Primary Care Clinic / Center| https://athena.ohdsi.org/search-terms/terms/38004247.

I've not seen anyone use this attribute in real life in fairness (most of the work I've been a part of doesn't find it relevant)...

Somewhat related to #81 https://github.com/OHDSI/dbt-synthea/pull/81

— Reply to this email directly, view it on GitHub https://github.com/OHDSI/dbt-synthea/issues/85, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAYR4GJYASJFPPOPOE5X5ADZ5AI7RAVCNFSM6AAAAABQPXDXAOVHI2DSMVQWIX3LMV43ASLTON2WKOZSGYYDSOBZGMZDMMY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

katy-sadowski commented 4 weeks ago

I like this idea too! It seems like the docs suggest inferring place of service from the types of visits associated with a care site - https://ohdsi.github.io/Themis/mapping_place_of_service_concept.html. Maybe we build some logic around this?

I don't love using the name of the organization to infer place of service because in my experience this name is usually the legal name of a business and could be misleading. On the other hand, looking at the aggregate of visit types to derive the place of service feels like it might be overly complicated and also not necessarily accurate either...

@heschmidt04 - I think most type concepts are set in the event tables but probably worth a re-review and confirmation we like the choices.