FHIR Chat · Microsoft PowerBI - FHIR Connector · analytics on FHIR

Stream: analytics on FHIR

Topic: Microsoft PowerBI - FHIR Connector


view this post on Zulip Paulo Rades (Nov 05 2019 at 19:27):

Hello.

Is Microsoft powerBI a good choice for working with FHIR? Is there a connector ready to consume server data?

Thanks for any help.

view this post on Zulip nicola (RIO/SS) (Nov 06 2019 at 14:57):

He - we usually "flatten" and export data from aidbox into PowerBI. For example using custom operation $sql-dump - https://docs.aidbox.app/basic-concepts/bulk-api-1/usddump-sql-tutorial

view this post on Zulip Josh Mandel (Nov 07 2019 at 14:44):

Re: connection directly to a fhir server, can you share some of these cases you have in mind? In general for analytics it can often be helpful to connect to the underlying database of a server (or an export -- as Nicola describes, or using the standard $export operation) rather than connecting through the fhir API itself (which works best for more targeted queries).

view this post on Zulip Rik Smithies (Nov 08 2019 at 23:47):

@Josh Mandel If it's a Microsoft server (the open source one) what are good options for querying outside of the normal REST API?
Is there a way to query Cosmos DB for instance?

view this post on Zulip Josh Mandel (Nov 08 2019 at 23:49):

@Michael Hansen for an authoritative answer (and I since he has also written Azure Data Factory scripts for extraction) -- but yes, I think there should be access to underlying storage.

view this post on Zulip Paulo Rades (Nov 11 2019 at 13:43):

Hello friends. Thanks for the answers....

A little more information about the project:

We have 9 different data sources (9 different applications that implement relational databases like MySql, Postgress, Oracle, Sql, etc.)

Challenge:

1 - create a structured data lake in FHIR
2 - enable information exchange of each of these applications with FHIR
3 - Analyze the data from this data lake we are building.

Doubts:

1 - Is it viable to use Aid.Box as suggested by Nicolas, since our original databases are relational (sql) ....?

Note: I want to use mirth connect to turn relational bds into FHIR .. I don't understand why move this data to Aid.Box and again have a relational database to perform the analysis?

2 - For each application should I build a FHIR server for information exchange or would a single FHIR server be the best option? Or would Aid.Box also be used for data exchange with other applications?

Thank you in advance

view this post on Zulip nicola (RIO/SS) (Nov 27 2019 at 05:16):

@Paulo Rades aidbox is just one of ready-to-use generic FHIR servers with extra features, it's up to you use it or not. The problem with FHIR and relational databases is that it's hard to fit FHIR in general into traditional tables (because object relational impedance mismatch). You have 9 applications and you want to aggregate data in one place? Can this 9 apps have unified relational structure? Usually not. That's why you probably want to load this data into generic FHIR server and after flatten it again (with some reasonable information loss) for analytics. From other side, if you able to find common relational denominator for all of your apps - you can skip this step.

view this post on Zulip Muhammad Abubakar Ikram (Jun 01 2020 at 07:50):

I have created a very basic open source FHIR Connector for Google Data Studio (it's free for use).
If you wish to use Google Data Studio you can start from here. https://github.com/leo9223/FHIR-GDS-Connector


Last updated: Apr 12 2022 at 19:14 UTC