I think I got the bulk of the work done for diagnoses
(#63<http://informatics.gpcnetwork.org/trac/Project/ticket/63>). It's loaded
into babel<http://babel.gpcnetwork.org/i2b2/webclient/> under GPC - Diagnoses.
The main differences from the KUMC HERON Diagnoses tree are
* It uses ICD9CM: rather than ICD9: as the scheme.
* It excludes the Epic DX codes that are finer than ICD9CM diagnoses.
* It includes the top "DISEASES AND INJURIES" folder; we flatten that in
HERON.
Here are the detail from GetTermInfo from dragging ALS (335.20) into a panel:
<concept>
<level>7</level>
<key>\\GPC_Diagnoses\i2b2\Diagnoses\A18090800\A8359006\A8359018\A8359773\A8344376\A8352659\335.20\</key>
<name>Amyotrophic lateral sclerosis</name>
<synonym_cd>N</synonym_cd>
<visualattributes>LA </visualattributes>
<totalnum xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance";
xsi:nil="true" />
<basecode>ICD9CM:335.20</basecode>
<facttablecolumn>concept_cd</facttablecolumn>
<tablename>concept_dimension</tablename>
<columnname>concept_path</columnname>
<columndatatype>T</columndatatype>
<operator>like</operator>
<dimcode>\i2b2\Diagnoses\A18090800\A8359006\A8359018\A8359773\A8344376\A8352659\335.20\</dimcode>
<tooltip>GPC - </tooltip>
</concept>
As you can see, the concept paths are computed (in
umls_dx_concepts.sql<https://informatics.kumc.edu/work/browser/heron_load/umls_dx_concepts.sql>*)
in a somewhat atypical fashion. I discovered UMLS provides a mrhier.ptr column
that works great as an i2b2 concept path, once you replace '.' by '\'.
The tooltips don't have the usual stuff; in HERON, we populate those for all
terms at once in a different part of the code. Does anybody particularly care
about those for shared GPC terminology purposes?
I wonder what should be the deliverables for (this part of) our shared
terminology?
* availability for exploratory use in the babel i2b2 web client
* a CSV file of i2b2 metadata
* SQL source code to create the i2b2 metadata from the results of running
the UMLS Oracle SQL load scripts
The first is done, provided the details above are acceptable to everybody.
The CSV file is on babel; I'd have to check on the licensing details for ICD9
to know whether I can post it publicly.
As to the SQL code, I tweaked umls_dx_concepts.sql and the python drivers
slightly today; the details aren't in that public copy, but only via the
MultiSiteDev<http://informatics.gpcnetwork.org/trac/Project/wiki/MultiSiteDev>
process (rev 3571c742ead5). There's more than can/should be done to disentangle
this code from other parts of HERON.
And fixing search by codes using ICD9CM: is still todo
(#59<http://informatics.gpcnetwork.org/trac/Project/ticket/59>).
--
Dan
_______________________________________________
Gpc-dev mailing list
[email protected]
http://listserv.kumc.edu/mailman/listinfo/gpc-dev
