ctakes-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Miller, Timothy" <Timothy.Mil...@childrens.harvard.edu>
Subject Re: cTAKES Sectionizer: how to integrate it with clinical pipeline - Solved
Date Fri, 06 Dec 2013 12:32:36 GMT
Glad you didn't encounter any issues. As far as getting it running there wasn't any particular
issue, my worry is just the 'unknown unknowns' as they say. I think for performance I have
one worry yet, especially if it is the default sectionizer -- does it fail gracefully and
will it ever skip text? I think it should be ok but that would be one thing it would be worth
testing. So, Paula, what happens if you change the spelling of a section header (i.e., introduce
a typo)? And just out of curiosity, what kind of notes are you running it on? Any particular
dataset?
Thanks
Tim

On 12/05/2013 09:04 PM, digital paula wrote:
Pei,

I appreciate you mentioning the preferredText feature for getting section headings to render,
the first column in the mapping  file should suffice.

In a previous post, Tim stated that the sectionizer would be a huge benefit to the research
community once it's working or something along those lines.   What was the problem with getting
it to work?  I ask because I didn't encounter any issues during my preliminary testing.  All
I did was an integration and minor configuration, as stated in my previous post.   The reason
why I'd like to know is so I'm cognizant of any known issues in case I encounter them once
I get back to using the sectionizer...should be in a few days.

Thanks.

Regards,
Paula

________________________________
From: Pei.Chen@childrens.harvard.edu<mailto:Pei.Chen@childrens.harvard.edu>
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: RE: cTAKES Sectionizer: how to integrate it with clinical pipeline - Solved
Date: Thu, 5 Dec 2013 14:26:56 +0000


Paula,

Glad to hear it’s working for you.  Please feel free to let us know how it works out for
you in your use case and dataset.



>I didn't see it in the trunk so I  manually added the text file ccda_sections.txt taken
from the sandbox.    I added it under src/main/resources in ctakes-core:

org/apache/ctakes/core/sections/ccda_sections.txt

Check out: http://svn.apache.org/r1547576 ctakes/trunk/ctakes-core-res/src/main/resources/org/apache/ctakes/core/sections/ccda_sections.txt
  (with props)



>This would be good to have too: segmentHeading:  HISTORY OF PRESENT ILLNESS:

There is a field called Segment.preferredText.  Which should display the first text column
in the mappings file…


Thanks,

Pei



From: digital paula [mailto:cybersation@hotmail.com]
Sent: Wednesday, December 04, 2013 8:59 PM
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: RE: cTAKES Sectionizer: how to integrate it with clinical pipeline - Solved



Pei,


Okay, the sectionizer is now integrated in the clinical pipeline and I did some preliminary
testing to confirm.  I added the CDASegmentAnnotator lines that you stated to the Aggregate
Descriptor and commented the simple annotator in the flow.  In addition, I also had to comment
out this in order for the Aggregate descriptor to save with no error:

<!--

<configurationParameter>
<name>SegmentID</name>
<description/>
<type>String</type>
<multiValued>false</multiValued>
<mandatory>false</mandatory>
<overrides>
<parameter>SimpleSegmentAnnotator/SegmentID</parameter>
</overrides>
</configurationParameter>
-->



I didn't see it in the trunk so I  manually added the text file ccda_sections.txt taken from
the sandbox.    I added it under src/main/resources in ctakes-core:

org/apache/ctakes/core/sections/ccda_sections.txt

I tested on a few narratives and I'm attaching what returned using CVD tool for one of them.
   The segmentID was populated with segmentID:  1.3.6.1.4.1.19376.1.5.3.1.3.4.     I looked
in the ccda_sections.txt file and this was it:



1.3.6.1.4.1.19376.1.5.3.1.3.4,10164-2,HISTORY OF PRESENT ILLNESS,brief history of physical
illness,history of present illness,history of the present illness



I looked back in the narrative and the heading was:  HISTORY OF PRESENT ILLNESS: The patient.....

I just tested on a few narratives and though that hardly constitutes  testing because I need
to resolve a more urgent issue for my research but will return to this for full testing entailing
the application to all of my narratives.   However, from a preliminary perspective, it looks
good....only thing I'd like to see is the actual text as well for the segment heading as another
feature.  For example what displays is this:
segmentID:  1.3.6.1.4.1.19376.1.5.3.1.3.4
This would be good to have too:
segmentHeading:  HISTORY OF PRESENT ILLNESS:

Thanks.


Regards,
Paula




> From: Pei.Chen@childrens.harvard.edu<mailto:Pei.Chen@childrens.harvard.edu>
> To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
> Subject: RE: cTAKES Sectionizer: how to integrate it with clinical pipeline
> Date: Tue, 3 Dec 2013 20:38:22 +0000
>
> Paula,
> I moved the sectionizer to trunk now and added the xml descriptor for it.
>
> In your Aggregate Descriptor, just add:
> <delegateAnalysisEngine key="CDASegmentAnnotator">
> <import location="../../../ctakes-core/desc/analysis_engine/CDASegmentAnnotator.xml"/>
> </delegateAnalysisEngine>
> .
> <node>CDASegmentAnnotator</node>
>
> If you would like to see it wired together via uimaFIT, check out the test case:
> ctakes-core/src/test/java/org/apache/ctakes/core/ae/TestCDASegmentAnnotator.java
>
> Hope that helps.
> It might be even worthwhile defaulting to this instead of the SimpleSegment (since simple
segment does nothing more than span the entire document...)
> --Pei
>
>
>
> From: digital paula [mailto:cybersation@hotmail.com]<mailto:[mailto:cybersation@hotmail.com]>
> Sent: Tuesday, December 03, 2013 1:51 PM
> To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
> Subject: cTAKES Sectionizer: how to integrate it with clinical pipeline
>
> Hi Pei,
>
> Last week we discussed briefly the sectionizer and now that I have it loaded successfully
I just need to integrate it in the clinical pipeline.
>
> The sectionizer doesn't have a desc folder with associated XML descriptor and I understand
that things are moving towards UIMAfit so that's probably the reason why it doesn't.  Can
you provide some guidance on what you'd recommend for testing the sectionizer?  That is, should
I just create the XML Descriptor using one of the reference materials from the UIMA website
for creating descriptors or would you recommend  using UIMAfit?  If the latter, can you provide
assistance of how to integrate the sectionizer into the AggregatePlaintext AE using UIMAfit.
>
> Thanks.
>
> Regards,
> Paula


Mime
View raw message