Stream: implementers
Topic: CQL stream
Erich Schulz (Jun 18 2016 at 00:08):
because I talk too much, and I'm way to excited, and its a big topic, I've been bold and created a CQL stream
Erich Schulz (Jun 18 2016 at 00:11):
co-conspirators are most welcome
Erich Schulz (Jun 20 2016 at 06:43):
especially co-conspirators with an interest in javascript and/or decision support and/or clinical quality indicators
Erich Schulz (Jun 20 2016 at 06:44):
the transpiler is starting to spit out plausbile looking javascript now
Erich Schulz (Jun 20 2016 at 06:45):
so far progress has exceeded my expectations but still a lot of work to do!
Erich Schulz (Jun 26 2016 at 02:07):
an update, i now have the basic (ie the key 20%) language operational, today I'm writing an a "mock" to allow CQL code to Retrieve FHIR data so i can get a basic CQL 'Query' operation running
Erich Schulz (Jun 26 2016 at 02:08):
I would love to hear from anyone interested in CQL who has some queries and testdata (ideally expressed in CQL and FHIR+JSON respectively)
Erich Schulz (Jun 26 2016 at 03:25):
also if there are any experienced FHIR folk around today who can give me a hand getting code and date types done in a standardish way, please see the "pointy end" thread in the CQL stream
Chris Moesel (Jul 06 2016 at 17:29):
It's great to see some movement around CQL and FHIR! Hopefully you've seen the CoffeeScript java execution project by now. It's not a transpiler, but rather a JSON ELM interpreter. It already handles a very large portion of the spec (I'd guess > 90%) and we have successfully used _browserify_ to compile it down to a single JS file that works in a browser. Since you reference the CQL-to-ELM compiler, I'm assuming you were already aware of it, but if not:
https://github.com/cqframework/clinical_quality_language/tree/master/Src/coffeescript/cql-execution
Within that tree you'll find a partial FHIR solution, but there are two caveats: (1) it uses an older version of FHIR, and (2) it currently expects flat files of bundles (each representing a patient). Within the "test/elm" folder, you'll find test cases that use FHIR resources as the test data to run against. These are unit tests, so they tend to be at the more detailed level (e.g., does the "multiply" operator work) rather than real use cases. But still, it may be helpful if you haven't seen it yet.
Erich Schulz (Jul 07 2016 at 13:05):
Thanks @Chris Moesel. Yes I am currently working through those test cases now and asking @Bryn Rhodes a series of daft noob questions. I am planning to establish a pattern based on promises that should enable use of restful fhir queries. To be honest I probably should have just gone with your coffee script interpreter but when I raised the desire for a compliler @Grahame Grieve egged me on to write one myself and I took the bait. I am only doing it in my spare time but progress is faster than I expected. I think the code scrutiny of the spec that I am giving will at least uncover glaring issues as well as lead me to have a good grounding in the language.
Chris Moesel (Jul 07 2016 at 13:16):
Thanks for the update, @Erich Schulz. I'm glad to hear that it's moving along more quickly than expected. Hopefully that means we did something right when we designed all of this stuff! We tried to strike some balance between ease of authoring and ease of implementation -- and I think ELM allowed us to hit that balance well (ease of authoring through CQL, ease of implementation through ELM). Definitely interested in your experience as you continue on! We (MITRE) did an HQMF-to-JS transpiler for the HQMF/QDM-based eCQM measures. While it worked, we found it difficult to debug the measures sometimes, but I think that was due more to some implementation-unfriendly features of QDM. I expect an ELM-to-JS transpiler should be cleaner -- and it will be interesting to see the pros and cons of the transpiler vs the interpreter. It's always good to have choices!
Grahame Grieve (Jul 07 2016 at 20:46):
I egged Erich on? phooowar....
Erich Schulz (Jul 09 2016 at 09:20):
I think the CQL pattern of shareable knowledge artefacts is very much way to go - we need to get more international interest in it!
Erich Schulz (Jul 09 2016 at 09:23):
@Grahame Grieve the internet never lies ;-)
Erich Schulz (Jul 09 2016 at 09:23):
mind you... if someone told me to jump off a cliff...
Erich Schulz (Jul 09 2016 at 09:24):
but CQL+FHIR is some serious awesomeness waiting to happen
Erich Schulz (Jul 09 2016 at 09:25):
wish i had more hours in day
Grahame Grieve (Jul 09 2016 at 11:41):
no, the internet never lies: https://xkcd.com/386/
Grahame Grieve (Jul 09 2016 at 11:41):
you can get more hours in the day if you sleep less.
Grahame Grieve (Jul 09 2016 at 11:42):
this works well up to a point, but then your outcomes deteriorate. The point at which this happens is variable, but most people have less thant optimal outcomes once sleep moves towards a negative number
David Hay (Jul 09 2016 at 21:49):
how does negative sleep work?
Erich Schulz (Jul 09 2016 at 22:42):
no sleep while doing above normal awakeness?
Brian Postlethwaite (Jul 10 2016 at 12:02):
That must be what you record for sleep you get on a plane from Los Angeles to Auckland?
Last updated: Apr 12 2022 at 19:14 UTC
