Feb 16

IntenseDebate Test (Updated)

[Update] One’s FriendFeed feed needs to be public in order for IntenseDebate to pick up the comments and bring them onto the blog. IntenseDebate isn’t perfect, but I love the threaded comments, I like quickly moderating comments via email, and I like the sidebar widget for comments (if you’re reading this via RSS, visit the blog and check out the left sidebar). I’ll keep it for now. Thanks to those who helped me test it![/Update]

____________

I’ve installed a plugin/service called IntenseDebate on this blog. Among the things it is supposed to do is pick up comments people make on my posts in FriendFeed and import them as comments. I’m curious to see if it will find those comments even if my FriendFeed is set to private. If you’re seeing this post through FriendFeed, please leave a comment in FriendFeed (not a ‘like’) so that can be tested?

Thanks!

Feb 12

Yale Image Finder (and UC Berkeley’s BioText)

The Yale Image Finder searches PubMed Central articles for images.


That sounded not only like a good idea, but a good idea I’d heard before. In July of 2007, I posted about UC Berkeley’s BioTextSee an article about BioText here, which seems to already search PubMed Central for images. Why build another tool to do the same thing?

The answer is found in this Bioinformatics article

The authors note they are aware of BioText, but that…

…we are not aware of a biomedical search engine that can retrieve images by searching the text within biomedical images. This offers several advantages over searching over captions alone. First, captions may not contain all the textual information that is contained in the images. Second, image texts are usually very specific, allowing for precise matching of images with related images.

Neat. So Yale’s tool does OCR on text appearing in images and adds that text to its searching index of images in PubMed Central. Cool.

Feb 11

WebPax.com

I’m not sure what to make of WebPax.com…but at first glance, it seems really cool to have a Web-based service for viewing images in DICOM format. I know at least a couple of physicians who will want to try it out right away for sharing the occasional scan with a colleague from a distance.

I *do* like that DICOM files are anonymized as they are uploaded. DICOM tags are cleared and…

• The year and month are not modified
• The day is set to the first of the month
• The time is set to midnight

The patient’s birth date is set to January 1, 1970

I’ll say this much: If I kept a digital personal health record in an online service, I’d want to be able to view DICOMs in it with this kind of tool. Google needs to buy these guys or build a comparable tool. Maybe that’s what they and IBM can work on next.

Feb 09

Directory of Librarians Who Twitter

Most know I’m not a huge fan of Twitter (I prefer FriendFeed), but this interests me anyway.

JustTweetIt is a service intended to help Twitter users find others they may want to follow and includes directories. I recently stumbled across JustTweetIt’s directory of Twittering Librarians

Check it out, see if there are any librarians listed you want to follow and consider adding an entry to help others find you.

Feb 08

HAVIDOL (avafyneyme HCI)

Dated 2007 but new to me:

Havidol is clearly an amazing new drug. Thank goodness there’s such a wonderfully detailed site to tell us all about Havidol and how it can treat Dysphoric Social Attention Consumption Deficit Anxiety Disorder (DSACDAD).

Click to visit the site

Click to visit the site

Great parody of direct-to-consumer advertising.

Feb 05

HHS/FDA/CDC Social Media Tools for Consumers and Partners

New to me- and a good idea to put all of this on one page.

http://www.cdc.gov/socialmedia/

I didn’t know the CDC was on MySpace or that the FDA had a recall Twitter feed.

I decided I should definitely follow the CDC’s Twitter feed for Health Professionals, which is for “…Health Professionals interested in staying up-to-date with CDC’s interactive media activities…”

They’ve also got a widget to help consumers search for products impacted by the Peanut-Containing Product Recall (embedded below).

FDA Salmonella Typhimurium Outbreak 2009. Flash Player 9 is required.

Includes:

  • Blogs
  • eMail Subscriptions
  • Health-e-Cards
  • Mobile Information
  • Online Video
  • Phone/Email
  • Podcasts
  • RSS Feeds
  • Social Networks
  • Badges for Social Networks
  • Twitter
  • Virtual Worlds
  • Web Sites
  • Widgets

Go check it out.

Hat tip: Maura Sostack

Feb 03

More on Evaluating Health Journalism

Francesca Frati (who rules) pointed out last week a site produced by the Royal College of Physicians of Edinburgh and Royal College of Physicians and Surgeons of Glasgow: http://behindthemedicalheadlines.com/.

Craig Stoltz (previously mentioned) dropped me an email to point out a post I’d missed from The Health Care Blog by Alicia White of Bazian (the company which evaluates stories for the NHS’s Behind the Headlines service).

Says Ms. White:

…we’ve developed the following questions to help you figure out which articles you’re going to believe, and which you’re not.

Questions include:

  • Does the article support its claims with scientific research?
  • Is the article based on a conference abstract?
  • Was the research in humans?
  • How many people did the research study include?
  • Did the study have a control group?
  • Who paid for and conducted the study?
  • Did the study actually assess what’s in the headline?
  • How can I find out more?

Good stuff. Go read it.

Thanks again for the pointers, Francesca and Craig!

Feb 02

novo|seek (3rd-Party PubMed/MEDLINE Tool)

(First started drafting this post on 10/27/2008, then posted an incomplete draft by accident on 2/2/2009. Sorry ’bout that.)

I received an email recently which invited me to try the beta version of novo|seek.

Novo|seek is the product of a Madrid-based company called bioalma (a.k.a. Alma Bioinformatics).

novo|seek is an information extraction system for searching the published knowledge in biomedical literature.

novo|seek index the biomedical literature with a text mining technology that enables identify uniquely the key biomedical terms. To do this unambigous identification the technology takes into account external available data and contextual term information. As a result of this indexing technology novo|seek is able to retrive every document where a term is mention no matter the synonym used and discard those documents where the term is use with an unwanted meaning.

So…that means better term mapping? I wonder how the mapping compares to ReleMed.

Since “HRT” is something we know PubMed has had trouble mapping in the past, we’ll try looking for “HRT” in both ReleMed and novo|seek.

ReleMed mapped “HRT” quite nicely:

novo|seek was unable to map “HRT” and produced no results.

Moving on.

With novo|seek you can:

  • Extract precise information for over 3 million key biomedical concepts, no matter whether they are diseases, drugs, chemicals compounds, symptoms or genes.
  • Retrive key biomedical concepts and bibliographic information to your query.
  • Highlight relevant biomedical concepts in the text.
  • Filter your results fast and easy.
  • Review key information derived from over thousands of documents in a single screen.
  • Search for an author and find key research concepts based on the analysis of his or her research.
  • Link to relevant external chemical and biological information.

I like the ability to filter by related concepts appearing in the left sidebar, but GoPubMed seems to do this in a similarly useful manner.

Click here for larger image

Click here for larger image

Click to see larger image

Click here for larger image

For that matter, PubMed PubReMiner does much the same thing.

Questions for novo|seek:

  • How does the relevance sorting work?
  • What does novo|seek do that none of the other 3rd-Party PubMed/MEDLINE tools do?
  • How does the mapping of search strings to “over 3 million key biomedical concepts” work?
  • Could you give an example where novo|seek’s mapping demonstrates superiority to PubMed’s or ReleMed’s?