You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ctakes.apache.org by Matthew Vita <ma...@gmail.com> on 2017/09/16 17:49:32 UTC
Followup Question on ICD10 Dictionary and Concept Matching Quality
Hi cTAKES Community, Sean, Tim,
As you recall, I recently put together the YouTube video "cTAKES: How to
Create an ICD10 Dictionary" (https://www.youtube.com/watch?v=4aOnafv-NQs)
to improve the documentation of using the NIH/UMLS and cTAKES tooling to
generate custom dictionaries. It is working well and Dr. Tim Miller's
cTAKES Docker repository will soon have further documentation/support for
using this approach in establishing multiple custom cTAKES dictionaries.
However, I am writing to because while everything with ICD10 is working,
I'm finding the concept matching to be inadequate as compared to
SNOMED/RXNORM (I'm sure I'm just missing something :) ). For example, if I
type in "Type 2 Diabetes", there is no concept match. However, when I type
"Type 2 Diabetes Mellitus", there is a match. Is there a way I can better
"train" or configure the dictionary to have matching at parity with that of
the SNOMED/RXNORM dictionaries?
For reference, here is the ICD10 configuration I'm using:
<dictionary>
<name>icd10Terms</name>
<implementationName>org.apache.ctakes.dictionary.lookup2.dictionary.UmlsJdbcRareWordDictionary</implementationName>
<properties>
<property key="jdbcDriver" value="org.hsqldb.jdbcDriver" />
<property key="jdbcUrl"
value="jdbc:hsqldb:file:resources/org/apache/ctakes/dictionary/lookup/fast/icd10/icd10"
/>
<property key="jdbcUser" value="sa" />
<property key="jdbcPass" value="" />
<property key="rareWordTable" value="cui_terms" />
<property key="umlsUrl" value="
https://uts-ws.nlm.nih.gov/restful/isValidUMLSUser" />
<property key="umlsVendor" value="NLM-6515182895" />
<property key="umlsUser" value="CHANGE_ME" />
<property key="umlsPass" value="CHANGE_ME" />
</properties>
</dictionary>
<conceptFactory>
<name>icd10Concepts</name>
<implementationName>org.apache.ctakes.dictionary.lookup2.concept.UmlsJdbcConceptFactory</implementationName>
<properties>
<property key="jdbcDriver" value="org.hsqldb.jdbcDriver" />
<property key="jdbcUrl"
value="jdbc:hsqldb:file:resources/org/apache/ctakes/dictionary/lookup/fast/icd10/icd10"
/>
<property key="jdbcUser" value="sa" />
<property key="jdbcPass" value="" />
<property key="umlsUrl" value="
https://uts-ws.nlm.nih.gov/restful/isValidUMLSUser" />
<property key="umlsVendor" value="NLM-6515182895" />
<property key="umlsUser" value="CHANGE_ME" />
<property key="umlsPass" value="CHANGE_ME" />
<property key="tuiTable" value="tui" />
<property key="prefTermTable" value="prefTerm" />
<property key="icd10cmTable" value="text" />
<property key="icd10amaeTable" value="text" />
<property key="icd10amTable" value="text" />
<property key="srcTable" value="text" />
<property key="icd10Table" value="text" />
<property key="icd10aeTable" value="text" />
<property key="icd10pcsTable" value="text" />
</properties>
</conceptFactory>
As always, I appreciate the hard work this community has done and, based on
the feedback from this thread, I will do my best to improve the
documentation for others.
Thanks,
Matthew Vita
www.matthewvita.com
RE: Followup Question on ICD10 Dictionary and Concept Matching
Quality [EXTERNAL]
Posted by "Finan, Sean" <Se...@childrens.harvard.edu>.
Hi Matthew,
Go to the directory that has your dictionary and run "grep -i "type 2 diabetes" <YourDictName>.script"
Do you see the line:
INSERT INTO CUI_TERMS VALUES(11860,2,3,'type 2 diabetes','diabetes')
?
If not, then it wasn't added when your dictionary was created. The dictionary will only contain terms from selected sources. The synonym "type 2 diabetes" is in the following sources:
"Type 2 Diabetes"
MEDLINEPLUS
MSH
NCI
NDFRT
"type 2 diabetes"
CHV
CSP
ICPC2P
MEDCIN
Make sure that when you install the umls you select one of those sources. If you only selected snomed as a source then you will not have the synonym that you want.
Sean
-----Original Message-----
From: Matthew Vita [mailto:matthewvita48@gmail.com]
Sent: Saturday, September 16, 2017 1:50 PM
To: dev@ctakes.apache.org
Subject: Followup Question on ICD10 Dictionary and Concept Matching Quality [EXTERNAL]
Hi cTAKES Community, Sean, Tim,
As you recall, I recently put together the YouTube video "cTAKES: How to Create an ICD10 Dictionary" (https://urldefense.proofpoint.com/v2/url?u=https-3A__www.youtube.com_watch-3Fv-3D4aOnafv-2DNQs&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=0rSQwwsnqghITPXgFr7F3NcE_8eV69HKjbyH_tkNlLg&s=s7WhFcIc-cj3SGUkzmi25-SxiPrf_LiVbu1sK9FhLE0&e= ) to improve the documentation of using the NIH/UMLS and cTAKES tooling to generate custom dictionaries. It is working well and Dr. Tim Miller's cTAKES Docker repository will soon have further documentation/support for using this approach in establishing multiple custom cTAKES dictionaries.
However, I am writing to because while everything with ICD10 is working, I'm finding the concept matching to be inadequate as compared to SNOMED/RXNORM (I'm sure I'm just missing something :) ). For example, if I type in "Type 2 Diabetes", there is no concept match. However, when I type "Type 2 Diabetes Mellitus", there is a match. Is there a way I can better "train" or configure the dictionary to have matching at parity with that of the SNOMED/RXNORM dictionaries?
For reference, here is the ICD10 configuration I'm using:
<dictionary>
<name>icd10Terms</name>
<implementationName>org.apache.ctakes.dictionary.lookup2.dictionary.UmlsJdbcRareWordDictionary</implementationName>
<properties>
<property key="jdbcDriver" value="org.hsqldb.jdbcDriver" /> <property key="jdbcUrl"
value="jdbc:hsqldb:file:resources/org/apache/ctakes/dictionary/lookup/fast/icd10/icd10"
/>
<property key="jdbcUser" value="sa" />
<property key="jdbcPass" value="" />
<property key="rareWordTable" value="cui_terms" /> <property key="umlsUrl" value="
https://urldefense.proofpoint.com/v2/url?u=https-3A__uts-2Dws.nlm.nih.gov_restful_isValidUMLSUser&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=0rSQwwsnqghITPXgFr7F3NcE_8eV69HKjbyH_tkNlLg&s=kxXUD2JtkzM0fU7rtE8DGCpCXTjF9ollYienRopcCF8&e= " /> <property key="umlsVendor" value="NLM-6515182895" /> <property key="umlsUser" value="CHANGE_ME" /> <property key="umlsPass" value="CHANGE_ME" />
</properties>
</dictionary>
<conceptFactory>
<name>icd10Concepts</name>
<implementationName>org.apache.ctakes.dictionary.lookup2.concept.UmlsJdbcConceptFactory</implementationName>
<properties>
<property key="jdbcDriver" value="org.hsqldb.jdbcDriver" /> <property key="jdbcUrl"
value="jdbc:hsqldb:file:resources/org/apache/ctakes/dictionary/lookup/fast/icd10/icd10"
/>
<property key="jdbcUser" value="sa" />
<property key="jdbcPass" value="" />
<property key="umlsUrl" value="
https://urldefense.proofpoint.com/v2/url?u=https-3A__uts-2Dws.nlm.nih.gov_restful_isValidUMLSUser&d=DwIBaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=fs67GvlGZstTpyIisCYNYmQCP6r0bcpKGd4f7d4gTao&m=0rSQwwsnqghITPXgFr7F3NcE_8eV69HKjbyH_tkNlLg&s=kxXUD2JtkzM0fU7rtE8DGCpCXTjF9ollYienRopcCF8&e= " /> <property key="umlsVendor" value="NLM-6515182895" /> <property key="umlsUser" value="CHANGE_ME" /> <property key="umlsPass" value="CHANGE_ME" /> <property key="tuiTable" value="tui" /> <property key="prefTermTable" value="prefTerm" /> <property key="icd10cmTable" value="text" /> <property key="icd10amaeTable" value="text" /> <property key="icd10amTable" value="text" /> <property key="srcTable" value="text" /> <property key="icd10Table" value="text" /> <property key="icd10aeTable" value="text" /> <property key="icd10pcsTable" value="text" />
</properties>
</conceptFactory>
As always, I appreciate the hard work this community has done and, based on the feedback from this thread, I will do my best to improve the documentation for others.
Thanks,
Matthew Vita
www.matthewvita.com