Advertisement

You Can't Google This: Flu Tracker Chokes on Big Data — Physician’s First Watch

Medical News |
March 14, 2014

You Can't Google This: Flu Tracker Chokes on Big Data

By Joe Elia

Google's flu tracker, meant to predict the extent of flu in the U.S. by tracking search terms, widely overestimated prevalence for the 2011–2012 and 2012–2013 seasons. An analysis of what went wrong and what it all means appears in Science.

The flu tracker, according to the analysts, "was an unstable reflection of the prevalence of the flu because of algorithm dynamics." The "most common explanation" (did they Google this?) was a media-stoked panic. In any event, the tracker erred on the high side — sometimes by as much as 50% — in 100 out of 108 consecutive weeks.

If you'll allow the speculation, couldn't another entity — one with access to all phone records — easily correlate increased calls to physicians' offices, and calls from those offices to pharmacies, and from those pharmacies to, say, Tamiflu distributors? That would give a reasonable picture of the flu's extent. Based only on metadata, of course.

Reader Comments (2)

Kathy Parker, MSN

Very sly! I had to go back and read it again. Thanks for keeping us entertained!

Mimi Thompson Breed MMS, ACC, Once Upon a Time

Way to go, Joe! The NSA as adjunct health care agency! They'll love it.

Your Comment

(will not be published)

Filtered HTML

  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Do you have any conflict of interest to disclose?
CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Image CAPTCHA
Enter the characters shown in the image.

Vertical Tabs

* Required

Reader comments are intended to encourage lively discussion of clinical topics with your peers in the medical community. We ask that you keep your remarks to a reasonable length, and we reserve the right to withhold publication of remarks that do not meet this standard.

PRIVACY: We will not use your email address, submitted for a comment, for any other purpose nor sell, rent, or share your e-mail address with any third parties. Please see our Privacy Policy.

Advertisement
Advertisement
Advertisement