ctakes-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From samir chabou <samir...@yahoo.com>
Subject Re: pico pipeline
Date Sun, 25 Aug 2013 17:05:48 GMT
Hi Pei, thanks for your feedback:
Yes there was some typo, I also  added some
cTakes annotators in the pico pipeline (see attached). 
 
Our purpose is: 
       1) To improve the
PICO recognition in abstracts text compared to what we are currently
testing with metamap
 
2 2) Through out my reading of some articles there
are some authors find that PICO is a useful organizing structure for clinical questions, otherthey
suggest it is less suitable for DiagnosisorPrognosis. From the few pico’s cases that I saw
(I
need to see a good sample of Pico’s Dignosis, Prognosis), I think :
 
a.       In the case of the Diagnosis and Prognosis the
focus of the question is most likely Output (O?)  (Need to be confirmed). Since the Output
is usually tend to be the relation between the Problem and Intervention (I -- O -->P or
C -- O --> P), consequently one of the thing to enhance Pico structure
for Prognosis and Dignosis is to enhance the aspect of relation recognition in
the text, to do so we are planning to write code/logic to add some relations
which will help the recognition of the O aspect.

b.       I also noticed that these
suggestions about pico on prognosis were advanced before 2006 where the focus
of the research was on the named entity recognition (NER) but few were done
about relation recognition.  I think
since then, the NLP techniques are evolved considerably  to be more efficient in relation
recognition
which will help our purpose.
 
 
Samir



________________________________
 From: "Chen, Pei" <Pei.Chen@childrens.harvard.edu>
To: samir chabou <samirchb@yahoo.com>; "dev@ctakes.apache.org" <dev@ctakes.apache.org>

Sent: Friday, August 23, 2013 10:41:46 AM
Subject: RE: pico pipeline
 

Hi Samir,
Perhaps others can chime in as well as I'm not too familiar with the proposed pipeline-
But it looks really interesting- especially the higher level components such as the Intervention
Annotator (are you planning to write code/logic to 'infer' what the cause or prognosis was?,etc).
Yes, one can certainly use the already annotated data from the existing components.  Some
notes:

-          SentenceDetector, Tokenizer is listed twice in the pipeline (is that just
a typo?)

-          I think there may be other components that you may want to include which might
help with your higher level annotators (some of which may or may not be available in mata-map):

o   Assertion (Negation, Subject, History-of, etc.)

o   Co Reference

o   Semantic Role Labeler

o   Temporal?
--Pei

From: samir chabou [mailto:samirchb@yahoo.com]
Sent: Thursday, August 22, 2013 11:43 AM
To: Chen, Pei; dev@ctakes.apache.org
Subject: pico pipeline

Hi Pei,
I'm trying to use ctakes to annotate PICO question concepts. I attached to you the pipeline
that I'll construct to do so. Please can you have a quick look to the attached file and tell
me if i'm on the right track or if you have some suggestions.
Thanks a lot
Samir
Mime
  • Unnamed multipart/mixed (inline, None, 0 bytes)
View raw message