Expanded code system preference lists #178
Conversation
| @@ -8,7 +8,7 @@ | |||
| class CodeDetectionBuilder(base_table_builder.BaseTableBuilder): | |||
| display_text = "Selecting unique code systems..." | |||
|
|
|||
| def _check_coding_against_db(code_source, schema, cursor): | |||
| def _check_coding_against_db(self, code_source, schema, cursor): | |||
There was a problem hiding this comment.
Need to have tests for this study - #179
| "urn:oid:1.2.840.114350.1.13.71.2.7.2", | ||
| "urn:oid:1.2.840.114350.1.13.71.2.7.4", |
There was a problem hiding this comment.
All these private/arbitrary systems need comments.
| @@ -21,7 +21,7 @@ CREATE TABLE {{ target_table }} AS ( | |||
| {%- endif %} | |||
| {%- if filter_priority %} | |||
| WHERE | |||
| u.codeable_concept.system = '{{ system }}' | |||
| u.codeable_concept.system LIKE '{{ system }}%' | |||
There was a problem hiding this comment.
Oh thats.. interesting. As a developer passing a system around, I would be very surprised to see it used as a prefix rather than a black-box string. http://example.org/animals/ being different from http://example.org/animals/lions for example. What's the use case for this - private urn namespaces? Is there maybe a way to opt-into the wildcard nature? But even then, I'm leery that we get a promise about all possible meanings / systems in the namespace?
There was a problem hiding this comment.
there's a lot of noise in the numeric ones specifically, and without the context, I couldn't determine what they corresponded to and how comprehensive they were, so rather than putting in 15 codes i tried to cheat it - i can flip this back and be more explicit.
Edit - this will probably have to stick to LIKE to handle the site-specific guid, but i'll remove the hard coded wildcard.
There was a problem hiding this comment.
ok, per our conversation offline, i've updated this to the explicit IDs, and removed the installation-specific guids from the cerner IDs.
I created #180 to track the need to think about some optimizations as these lists get larger.
dogversioning
force-pushed
the
mg/code_detection_update
branch
from
caf710c to
53c7930
Compare
| @@ -54,6 +54,12 @@ def denormalize_codes(self): | |||
| "http://snomed.info/sct", | |||
| "http://hl7.org/fhir/sid/icd-10-cm", | |||
| "http://hl7.org/fhir/sid/icd-9-cm", | |||
| "http://hl7.org/fhir/sid/icd-9-cm/diagnosis", | |||
| # Cerner specific systems | |||
There was a problem hiding this comment.
That doesn't seem entirely true - it's a spec standard way of indicating why data is gone. Epic doesn't use it though, it sounds like.
| "http://hl7.org/fhir/sid/icd-9-cm/diagnosis", | ||
| # Cerner specific systems | ||
| "http://terminology.hl7.org/CodeSystem/data-absent-reason", | ||
| # EPIC specific systems |
There was a problem hiding this comment.
I'm guessing we can't say anything else about them? But it'd be nice if we could distinguish what these two mean even a little bit.
There was a problem hiding this comment.
yeah, hard agree!
| "http://hl7.org/fhir/sid/icd-10-cm", | ||
| "http://hl7.org/fhir/sid/icd-9-cm", | ||
| # Cerner specific systems | ||
| "https://fhir.cerner.com/%/nomenclature", |
This PR makes the following change:
Checklist
docs/) needs to be updated