Hello, Thanks Sean, always helpful - will be able to test soon and update. (I have a couple questions on disambiguation but will create another thread)
Kind Regards, Eugenia Monogyiou | NTT Data UK Consulting & IT Solutions Ltd. 1 Royal Exchange, London EC3V 3DG Mob: +44 (0)7971623683 Email: eugenia.monogy...@nttdata.com -----Original Message----- From: Finan, Sean <sean.fi...@childrens.harvard.edu> Sent: 02 December 2020 14:56 To: dev@ctakes.apache.org Subject: Re: Lab entity and Temporal attributes related enquiries [EXTERNAL] Hi Eugenia, >1. - A few years ago Kean Kaufmann wrote a nice annotator that can discover lab values. https://issues.apache.org/jira/browse/CTAKES-441 It is in ctakes as org.apache.ctakes.core.ae.LabValueFinder It may not be exactly what you are looking for, but you can see from the comments in jira that TUIs may be used to fetch lab annotations. >2. I am definitely not the expert on the temporal module, so somebody else may write me wrong ... > I was looking into the "event-time" features I think that is a good route to take. I would take a look at the CalendarUtil class in org.apache.ctakes.core.util You may be able to use the method Calendar getCalendar( Annotation ) or Calendar getCalendar( String ) Use the TemporalTextRelations, looking for those relating an Event with a Timex. Then call the getCalendar method with the Timex. Note that the relation will have a category indicating relativity such as the event being before, after, or overlapping the time. Though I have never used it, there is some time normalization in a ctakes-temporal ae named WithinSentenceBeforeRelationAnnotator. I think that Steven Bethard contributed the code there and in the ctakes-temporal Utils class to integrate with his timenorm project https://github.com/clulab/timenorm You can also take a look at lines #145 - #155 in the class org.apache.ctakes.temporal.utils.CalendarUtil Calendar getCalendar( String ) You will see that the method uses a 3rd party library named jChronic https://github.com/samtingleff/jchronic Chronic.parse( String ) returns a Span object that contains a normalized start and end date. This can be useful for spans of time in addition to points in time. Sean ________________________________________ From: Monogyiou, Eugenia <eugenia.monogy...@nttdata.com> Sent: Wednesday, December 2, 2020 8:32 AM To: dev@ctakes.apache.org Subject: Lab entity and Temporal attributes related enquiries [EXTERNAL] * External Email - Caution * Hello and happy Wednesday, Two independent functionality topics this time :) 1. Labs : As I understand "Labs" in the refSem namespace is a separate core clinical element from "Procedure" but I haven't managed to get any Lab annotations regardless of which pipeline I am using and I have tried a few. Is there a specific NER component or configuration that produces such annotations? 2. Temporal attributes : I have included the TemporalSubPipe from the trunk in my custom pipeline and I do get the "DocTimeRel" features but the "startTime" and "endTime" event attributes are never populated. I am interested in retrieving the actual date associated with a prescription or diagnosis in a clinical letter hence I was looking into the "event-time" features rather than "event-event" features but I am not sure whether I am looking in the right place or have I even identified the right attributes, properties etc. Also, I am not sure whether this retrieval would require re-training of a model or with code. Any ideas/ guidance would be greatly appreciated! Many thanks in advance Kind Regards, Eugenia Disclaimer: This email and any attachments are sent in strictest confidence for the sole use of the addressee and may contain legally privileged, confidential, and proprietary data. If you are not the intended recipient, please advise the sender by replying promptly to this email and then delete and destroy this email and any attachments without any further use, copying or forwarding. Disclaimer: This email and any attachments are sent in strictest confidence for the sole use of the addressee and may contain legally privileged, confidential, and proprietary data. If you are not the intended recipient, please advise the sender by replying promptly to this email and then delete and destroy this email and any attachments without any further use, copying or forwarding.
