How to integrate FHIR Engine with SQL Server

[Vishal Parekh]

Hi,

I am a .net developer, trying to build an API to allow third parties to get clinical data as per FHIR standard. I download and implement SPARK Server (https://github.com/furore-fhir/spark). It works as expected with mongo database. But I have to implement the same with SQL Server, as that, we are using as a database in our existing EMR application.

I am new with asp.net, so does not understand how to integrate SQL Server in place of mongo.

Any help will be appreciated.

[magnus....@gmail.com]

Hi,

I have seen a couple of discussions on this topic earlier. Hopefully these links can point you in the right direction.

https://github.com/furore-fhir/spark/issues/94

https://github.com/furore-fhir/spark/wiki/Fitting-Spark-in-my-environment#using-spark-directly-against-your-own-datastore

https://github.com/furore-fhir/spark/issues/74

[Vishal Parekh]

Hi,

Thanks for your prompt reply.

I am trying to implement "SqlFhirStore" and "FhirDbContext". But somehow I am confused with implementation and development part.

So please guide me how to implement the same. 

Your help will be appreciated.

[magnus....@gmail.com]

My impression after reading on the subject and playing around with the current Spark is that making it work with SQL server would require a lot of work. The wiki say:

"...A search must be translated to a (usually SQL-) query into your own datastore. Therefore, supporting many search parameters will probably require a lot of work..."

If you do manage to get a functional SQL provider for Spark - please share it, i would really like one to.

[Vishal Parekh]

Sure,

But right now i am confuse whether i have to implement SQLFhirStore in Spark. Or i have to start writing my new server application with SQL Server as database. As our whole EMR system is developed with SQL Server, It is a bit difficult to migrate to mongo.

[Christiaan Knaap]

Connecting Spark to a (relational) database with an existing schema is even more difficult. It can be done, but you'd effectively be rewriting at least half of the server.
It is one of the features that we are enabling in Spark's successor Vonk, so that's why I know...

Christiaan Knaap
Lead developer Spark & Vonk