ctakes-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chen, Pei" <Pei.C...@childrens.harvard.edu>
Subject RE: Question pertaining to HistoryOf and SegmentID features
Date Wed, 27 Nov 2013 14:42:31 GMT
Paula,
It's already fixed in trunk :); feel free to give it a test.
https://issues.apache.org/jira/browse/CTAKES-274
Thanks for pointing that out- would you like to try out the sectionizer?  Just curious on
the use case for the sections.  Is it clinical research on top of plaintext notes?
--Pei

From: digital paula [mailto:cybersation@hotmail.com]
Sent: Wednesday, November 27, 2013 12:46 AM
To: user@ctakes.apache.org
Subject: RE: Question pertaining to HistoryOf and SegmentID features

Pei,

Thanks so much for the super prompt response.   A good reminder to us all.   When it's not
working as expected then try a clean/rebuild to fix it.   After the clean/rebuild it's working
as expected.

By the way,  as for your side note on whether it's worth updating the StatusAnnotator to populate
historyOf.    If the Assertion module is going to be released soon with historyOf feature
then I don't think it matters with updating Status Annotator historyOf feature since the Assertion
component really is a replacement for the Status Annotator.   However, if it's a minor effort
to make the Status Annotator populate historyOf, why not.    I'd like to fix it as my first
code contribution to the open source community.  :-)

Regards,
Paula
________________________________
Date: Tue, 26 Nov 2013 23:04:05 -0500
Subject: Re: Question pertaining to HistoryOf and SegmentID features
From: chenpei@apache.org<mailto:chenpei@apache.org>
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Paula,
Could you attach the desc xml and the output xml? Also, are  you using trunk?
Strange, i'm seeing uncertainty being set to 1/history:
<..DiseaseDisorderMention _indexed="1" _id="299" _ref_sofa="6" begin="11" end="19" id="5"
_ref_ontologyConceptArr="293" typeID="2" segmentID="SIMPLE_SEGMENT" discoveryTechnique="1"
confidence="0.0" polarity="0" uncertainty="1" conditional="false" generic="false" historyOf="0"
_ref_Segment="20"/>

Side note- Not sure if its worth updating the StatusAnnotator to populate the hisoryOf attribute
now (obviously those types were not avail previously.)

On Tue, Nov 26, 2013 at 10:37 PM, Pei Chen <chenpei@apache.org<mailto:chenpei@apache.org>>
wrote:
That annotator will set the IdentifiedAnnotation.*uncertainty()* with one of the following
values:
HISTORY_STATUS = 1;
FAMILY_HISTORY_STATUS = 2;
PROBABLE_STATUS = 3;

On Tue, Nov 26, 2013 at 10:32 PM, digital paula <cybersation@hotmail.com<mailto:cybersation@hotmail.com>>
wrote:
Hi Pei,

Thanks for the prompt response.    Okay, I uncommented out StatusAnnotator in the AggregatePlaintextPipeline
config/descriptor and tried again with this text.

"history of diabetes. hx diabetes. prior diabetes. "

Uncertainty and HistoryOf features remained 0.....doesn't look like anything happened.   
Note that I'm validating using the CVD debugger tool.

I did read something on that the Status and Assertion have overlapping functionality so I
commented out Assertion and still the same thing.    I looked at the StatusIndicatorFSM.java
file, text with history "hx" and "history" should be annotated so I'm not sure why it's not
working.

Am I missing some config?

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: Question pertaining to HistoryOf and SegmentID features
Date: Tue, 26 Nov 2013 20:46:12 +0000

Paula,
If you would like to use the old status annotator for history in the interim while the new
assertion module models are added.
One can just uncomment the below from your AggregatePlaintextPipeline config/descriptor:
        <!--
              <node>StatusAnnotator</node>
         -->
That annotator will set the IdentifiedAnnotation.uncertainty() with one of the following values:
HISTORY_STATUS = 1;
FAMILY_HISTORY_STATUS = 2;
PROBABLE_STATUS = 3;

--Pei

From: digital paula [mailto:cybersation@hotmail.com<mailto:cybersation@hotmail.com>]
Sent: Tuesday, November 26, 2013 10:00 AM
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: RE: Question pertaining to HistoryOf and SegmentID features

Hi Tim,

I remember you, you helped me out a while back!   I can't tell you how much that BagOfAnnotations.java
file has been  a Godsend!   Thanks again for posting it last month.

As for historyOf feature,   I don't so much mind performance so if you have something that
works for the historyOf with just only a performance issue, I'll still take it. I do appreciate
your prompt response on the historyOf feature which saved me from stepping through the code
which I've done a few times....the time spent  stepping through is equivalent to watching
a feature film and there's a lot going on in the code that I don't understand yet.

As for sementID's with the rule-based sectionizer/segmenter,  I'd like to help out with testing.
    It appears that the first step is manual configuration per your email.   Pei, when you
have a moment I'd like to  start with explanation of how it works then we can move forward
from there.

Thanks.

Regards,
Paula
________________________________
Date: Tue, 26 Nov 2013 09:24:20 -0500
From: timothy.miller@childrens.harvard.edu<mailto:timothy.miller@childrens.harvard.edu>
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: Re: Question pertaining to HistoryOf and SegmentID features

Hi Paula,
The historyOf module is currently under development, that explains those files you're seeing.
As of right now we are not satisfied enough with the performance to release it publicly. It
is marked 0 because it is defined as an integer field I believe, so it can't be assigned null.

As for the segment IDs in your other question, the default segmenter just makes the whole
document one segment because we don't yet have a module robust enough to run on notes from
any institution. We have been hoping to get one developed but nothing has materialized so
far. There is a rule-based sectionizer/segmenter in the cTAKES distribution but it requires
some manual configuration and is not very widely tested. If you are interested in helping
test it out that would be great. Pei would be the best person to explain to you how it works,
we'll see if he's able to chime in.
(Ah, I just checked and saw he already pointed you to it. If you want to try it we can help
you get it working.)

Thanks
Tim

On 11/25/2013 11:13 PM, digital paula wrote:
Matt, sorry but I sent the email by accident and didn't finish my thought.  Since there's
this file (historyOf.psv) in the Assertion folder it gives the impression that Assertion can
be configured to account for historyOf.   Though I tried some of the words such as "had diabetes"
or "prior diabetes" and value still 0 for historyOf feature.  I think it is configurable because
of the fact the default value for historyOf is 0 as opposed to <NULL> .  Maybe I should
step through the code and see where and when it's assigning 0.

Regards,
Paula

________________________________
From: cybersation@hotmail.com<mailto:cybersation@hotmail.com>
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: RE: Question pertaining to HistoryOf and SegmentID features
Date: Mon, 25 Nov 2013 23:09:33 -0500
Hi Matt,

it's me again.  Are you sure that historyOf feature can't be configured using Assertion. 
 I was looking around the Assertion folder and came across a file called historyOf.psv.  What
is this?

It contains the following:

has been|historyOf|historyOf

had|historyOf|historyOf

hx of recent|historyOf|historyOf

hx of|historyOf|historyOf

hx|historyOf|historyOf

recent|historyOf|historyOf

s/p|historyOf|historyOf

status post|historyOf|historyOf

post|historyOf|historyOf

previous|historyOf|historyOf

prior|historyOf|historyOf


________________________________
From: mcoarr@mitre.org<mailto:mcoarr@mitre.org>
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: Re: Question pertaining to HistoryOf and SegmentID features
Date: Tue, 26 Nov 2013 02:50:59 +0000
The ctakes assertion module does not have an option to set the historyOf attribute.  The assertion
module only sets subject, polarity, uncertainty, conditional, and generic.

Matt

From: digital paula <cybersation@hotmail.com<mailto:cybersation@hotmail.com>>
Reply-To: "user@ctakes.apache.org<mailto:user@ctakes.apache.org>" <user@ctakes.apache.org<mailto:user@ctakes.apache.org>>
Date: Monday, 2013-11-25 15:13
To: "user@ctakes.apache.org<mailto:user@ctakes.apache.org>" <user@ctakes.apache.org<mailto:user@ctakes.apache.org>>
Subject: RE: Question pertaining to HistoryOf and SegmentID features





Mime
View raw message