I'm looking for a tool to suppress sensitive information (e.g., HIV status, etc.) from free text clinical notes prior to allowing the notes to be published...
1917
Maury Pepper
mlpcmsstlcom
Mar 2, 2007 6:36 pm
<can-of-worms>; 1. How good does it have to be? Is 5% leakage of "sensitive information" OK? 2. Another view: ALL of the information is "sensitive". 3. Another...
1918
Fred Trotter
tr0tt3r
Mar 2, 2007 7:28 pm
Will, I am confused too. Wouldnt such a technology have to be turning test capable? Are you looking for something that can search Free Text make a ...
1919
Stuart Turner
stuartturner...
Mar 2, 2007 7:50 pm
... Will: In general, this falls under a number of natural language processing tools and specific steps toward tokenizing, chunking, part-of-speech tagging,...
1920
80n
b22gate
Mar 3, 2007 1:11 am
Will The only acceptable answer would be Maury's option 3. The patient decides. Anything else would be be inappropriate. And not just HIV status. The...
1921
David Forslund
dwforslund
Mar 3, 2007 1:23 am
Agreed. Such an action would at least have to have approval of a local HIPAA board. How would one "prove" it is reliable at removing protected information? If...
1922
Will Ross
wdross3
Mar 3, 2007 1:48 am
Dear 80n, This is, in fact, the use case in discussion. Assume the patient has agreed to suppress "detail x" from circulation beyond his/her physician's eyes...
1923
David Forslund
dwforslund
Mar 3, 2007 4:52 am
That is the reason for the patient consent in the HIPAA regulations. In my opinion, the patient would need to review the data to approve its release. The...
1924
Nandalal Gunaratne
nandalalx
Mar 3, 2007 9:38 am
Will, It is not a good idea to have sensitive information in free text. If you do, it should not go to "general circulation", right? How can one extract such...
1925
Nandalal Gunaratne
nandalalx
Mar 3, 2007 10:05 am
Thanks Ross! Due to your question i have come to know the present state of text mining and NLP. These will give you your solution I guess. ...
1926
Adrian Midgley
amidgley2@...
Mar 3, 2007 11:08 am
... As automation ... I think this is regarded everywhere as a hard problem. In the UK I believe it would be generally regarded as if not actually impossible,...
1927
David Forslund
dwforslund
Mar 3, 2007 4:25 pm
If one uses a structured report along the lines of the ASTM CCR, then I think it would be "relatively" easy to remove the sensitive information, since all of...
1928
Will Ross
wdross3
Mar 3, 2007 6:34 pm
Nandalal, I believe you are exactly right. In the case of this interesting problem the key issue is to identify the appropriate middleware services to safely...
1929
madhusasidhar@...
sasidhardoc
Mar 4, 2007 4:01 am
You may find the patient de-identifcation features of this project interesting: http://www.mii.ucla.edu/index.php/MainSite:NLPHome ... From:...
1930
JohnLeo Zimmer, MD
johnleoz
Mar 4, 2007 9:07 pm
... Will, Could you give a little more information on the structure of such a clinic? I am not aware of any practices with a "protected, physician-only area"...
1931
Will Ross
wdross3
Mar 5, 2007 2:35 am
John, In this use case, the record is not held by the health care site where the patient's confidential information is secure and private, but in a community...
1932
JohnLeo Zimmer, MD
johnleoz
Mar 5, 2007 3:33 am
... Thank you for your response, Will. I fear I am being intentionally obtuse. But I get stuck at this distinction between routine access and sensitive...
1933
Tim Cook
tw_cook
Mar 5, 2007 1:35 pm
Will, ... I believe what John is trying to explain is that ALL patient health information (PHI) is considered sensitive. Only the patient can determine which...
1934
John Norris
johnnojohnno
Mar 5, 2007 8:34 pm
In case you can't find anything, there are some tools to help you roll your own find-and-replace. I understand that may not get to removing certain semantics....
1935
Tim Churches
frednurkfred...
Mar 5, 2007 8:36 pm
See http://www.informationweek.com/news/showArticle.jhtml?articleID=197700935 and http://www.medsphere.com/press/20070226b Is that now a correct summation?...
1936
Joseph Dal Molin
dalmolin17
Mar 5, 2007 11:02 pm
Tim, The short answer is WorldVistA has some house cleaning to do....the website you pointed to is our old Sourceforge hosted site which is redirected to...
1937
Tim Churches
frednurkfred...
Mar 5, 2007 11:34 pm
... OK, thanks, that makes it much clearer. But is it fair to say that Medsphere has now satisfactorily answered the previous fairly vehement criticisms on...
1938
Fred Trotter
tr0tt3r
Mar 6, 2007 3:43 am
They have taken a step in the right direction, but they are not done. ... -- Fred Trotter http://www.fredtrotter.com [Non-text portions of this message have...
1939
Tim Churches
frednurkfred...
Mar 6, 2007 3:58 am
... Fred, What else do they need to do, in your opinion? Tim C...
1940
Fred Trotter
tr0tt3r
Mar 6, 2007 4:48 am
I have outlined what Medsphere needs to do to reconcile with the community here... http://www.gplmedicine.org/index.php?module=htmlpages&func=display&pid=13 In...
1941
Tim Churches
frednurkfred...
Mar 6, 2007 5:01 am
... What do you mean by "original licenses"? I thought that the VistA code on which Medsphere's OpenVista products are based was in the public domain, and thus...
1942
Fred Trotter
tr0tt3r
Mar 6, 2007 5:17 am
Tim, I suggest you wade through the mess under the blog post entitled "Medsphere betrays community" on GPLmedicine.org. This is a very complicated situation...
1943
Gregory Woodhouse
gwoodhouse
Mar 6, 2007 5:37 am
... I don't see how it clarifies your reference to the "original license". VistA is in the public domain in the sense that it may be obtained though FOIA....
1944
Tim Churches
frednurkfred...
Mar 6, 2007 5:38 am
... The issue is whether Medsphere is fit to be listed as a provider of open source health software on the OSHCA web site. I did wade through your blog entry,...
1945
JohnLeo Zimmer, MD
johnleoz
Mar 6, 2007 12:51 pm
... Thus, Medsphere's GPL license can be applied only to the <<differences>> between OpenVista server and FOIA VistA. (Likewise, any modifications World Vista...
