Stream: implementers
Topic: Data warehouse vs FHIR
Soren Klelund Borg (Feb 12 2022 at 19:48):
Hi everyone,
I have build, what I would call, classic data warehouses based on Kimball best practices on Microsoft technology for some time now. Normally it would look like this: Ingesting data to a sql server db or azure sql db -> transform data to one/more dimensional model -> tabular model on top -> PowerBI reports.
Now, I have been tasked with building a report based on (robotic) surgery data, cost data and more. "They" want it to be build on international standards without going in to detail what that means.
I have been looking at FHIR but I am having a hard time to figure out how it fits fit with a "classic" reporting solution build on top of a dimensional model. Is it just a matter of terminology (naming dimensions in a certain manner)? Please help me understand if FHIR could be of use in my case or send me in the right direction.
Thanks,
Søren
Lloyd McKenzie (Feb 12 2022 at 22:08):
FHIR data is not flat or naturally hierarchical. You can transform portions of it (e.g. with GraphQL) if you know both what kind of data you'll get and what sort of 'flatter' model will be appropriate for data analysis. FHIR is definitely not designed to do that natively though. You might find some helpful guidance in #storage for FHIR
Soren Klelund Borg (Feb 14 2022 at 12:38):
Lloyd McKenzie said:
FHIR data is not flat or naturally hierarchical. You can transform portions of it (e.g. with GraphQL) if you know both what kind of data you'll get and what sort of 'flatter' model will be appropriate for data analysis. FHIR is definitely not designed to do that natively though. You might find some helpful guidance in #storage for FHIR
THank you for your answer. So it is not really applicable in terms of data warehousing or how should this this be understood?
Grahame Grieve (Feb 14 2022 at 12:52):
see http://www.healthintersections.com.au/?p=2776
Soren Klelund Borg (Feb 16 2022 at 12:36):
Thank you both
Last updated: Apr 12 2022 at 19:14 UTC
