ctakes-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Masanz, James J." <Masanz.Ja...@mayo.edu>
Subject RE: Hi..
Date Mon, 24 Mar 2014 00:43:22 GMT
Last time I looked at that code, changing (adding a new property like "genesTuis") required
coding changes.
The expected property names are in UmlsToSnomedConsumerImpl.java and are used in if-else-if
It would be great to have that generalized someday, but for now, the options I know of, are
to modify that class, or to re-purpose one of the defined groups.

- James

From: Miller, Timothy [mailto:Timothy.Miller@childrens.harvard.edu]
Sent: Friday, March 21, 2014 8:52 AM
To: user@ctakes.apache.org
Subject: RE: Hi..

cTAKES by default gets the semantic groups/types you mentioned -- diseases/disorders, signs/symptoms,
and a few others that are clinically most relevant. Proteins/genes/etc. I think can be tagged
by editing the dictionary descriptor file, where you can specify the UMLS Type IDs (TUIs)
which you want included.

If you look in:

you will find lines like this:
                    <property key="anatomicalSiteTuis" value="T021,T022,T023,T024,T025,T026,T029,T030"/>
                    <property key="procedureTuis" value="T059,T060,T061"/>

which specify the TUIs ctakes will find. If you add the tui T116 for "amino acid,peptide,
or protein" I think you should get immunoglobulins. You can use the UTS to find the other
TUIs you need.

One issue remains, which hopefully someone else can address -- if her TUIs don't fit into
one of the standard ctakes semantic types (anatomical sites/procedure/ss/dd/drug), how does
she go about adding another and getting the type right? This is probably something we need
to be aware of as genetic information becomes more prevalent in clinical notes.

From: Prasanna Bala [balkiprasanna1984@gmail.com]
Sent: Friday, March 21, 2014 9:36 AM
To: user@ctakes.apache.org<mailto:user@ctakes.apache.org>
Subject: Hi..

I have some clarifications. I am able to run the tagger using UMLS for identifying disease,
drugs, signs/symptoms. I am afraid that I am not able to tag lot of good information. For
eg: Let me explain with an example. I tried documents that contains clinical test with lot
of medical entities in it. But I am not able to identify the entities such as  IMMUNOGLOBULIN,
TRANSGLUTAMINASE which should be tagged as protein or atleast chemical class. I tried the
same document for other biomedical text mining libraries such as ABNER. I am able to tag them
as protein. Am I missing something here ? Can you please suggest some modules for finding
genes, proteins, orgranism. What are the limitation of cTakes. Looking forward for someone
to clarify this.


View raw message