Stream: cds hooks
Topic: open source samples
Jens Villadsen (Aug 03 2017 at 11:30):
Are there any example CDS service implementations available as open source in java/scala/ruby?
Jens Villadsen (Aug 03 2017 at 11:44):
nevermind ... found what I was looking for @ http://editor.swagger.io/?url=https://raw.githubusercontent.com/cds-hooks/api/master/cds-hooks.yaml
Kevin Shekleton (Aug 03 2017 at 18:09):
@Jens Villadsen - Just a heads up that I need to update the Swagger API. We've made some recent changes in preparation for 1.0 that haven't yet been reflected in the Swagger API. Look for those changes soon
Jens Villadsen (Aug 03 2017 at 21:50):
(deleted)
Jens Villadsen (Aug 03 2017 at 21:51):
@Kevin Shekleton whats the ETA?
Kevin Shekleton (Aug 03 2017 at 21:54):
I'll have a PR submitted later today for this (working on it now)
Kevin Shekleton (Aug 03 2017 at 22:17):
@Jens Villadsen - Here you go: https://github.com/cds-hooks/api/pull/1
Jens Villadsen (Aug 15 2017 at 11:11):
Would it be considered against the CDS-standard to define such rich prefetch templates that all patient data is basically anonymous but still complete enough for a given CDS service to do its suggestion calculations?
Jens Villadsen (Aug 15 2017 at 11:11):
basically leaving out the patient ID and name
Jens Villadsen (Aug 18 2017 at 06:36):
@Kevin Shekleton any take on that?
Kevin Shekleton (Aug 18 2017 at 22:10):
@Jens Villadsen - It sounds like you are wanting to specify a prefetch template that pulls FHIR data across multiple patients. This is the first I've heard of this use case (everyone else has just been defining prefetch templates that are correlated to the current patient). With that being said, I don't see this going against the CDS Hooks spec.
Kevin Shekleton (Aug 18 2017 at 22:13):
I am curious though - is this aggregate data that your CDS service is requesting to be prefetched something that should be pull on every invocation of your service or something that you calculate separately?
One of the primary intentions behind prefetch was for the EHR to more efficiently give requested data to the CDS Service on the request. Eg, the EHR already has the requested data in memory or multiple CDS Services want the same data so the EHR can just grab it once and give it to each of the CDS Services requesting it
Kevin Shekleton (Aug 18 2017 at 22:14):
I know of another group that wants to use prefetch as a way for the EHR to send all data the CDS Service needs to perform its CDS as their network deployment does not allow the CDS Service to call back via FHIR to get any additional data
Jens Villadsen (Aug 21 2017 at 06:32):
that is also my case ... basically, I cannot actually assume that the service requesting CDS is really all that familiar with FHIR. While that may sound wierd, the case is that I need to support 'CDS' (as a term) in some kind of way, and the CDS standard seems like a proper way to go.
Jens Villadsen (Aug 21 2017 at 06:36):
The case is actually that I cannot even assume that the client invoking it knows much about FHIR. From a client perspective, it will probably seem proprietary. That also means that the prefetch templates will be more or less frozen and not be able to be expressed as dynamically as one could wish for.
Jens Villadsen (Aug 21 2017 at 06:39):
@Kevin Shekleton The intention is not to pull data across multiple patients at the same time. The intention is to be able to deliver decision support based on clinical recordings, e.g. weight, pulse, pregnancy, medication and so on without patient identification.
Last updated: Apr 12 2022 at 19:14 UTC
