Google Retools Its Flu Prediction Engine After Getting It Wrong

http://blogs.wsj.com/digits/2014/10/31/google-retools-its-flu-prediction-engine-after-getting-it-wrong/

by jakek • 12 years ago

40 3 12 years ago

blogs.wsj.com

refurb• 12 years ago

I had a chance to talk to a guy who worked in the pharmaceutical industry and it's amazing how much effort is put into predicting the flu season. Hundreds of millions of dollars are on the line not just predicting how many folks will get the flu, but also what strains will be the issue.

rtkwe• 12 years ago

Trying to predict the strains makes perfect sense, the flu shot is only ever effective against the specific spread of strains that it's created against. I presume there's some fall off of effectiveness or large increase in cost associated with creating a flu shot that targets more strains.

ggchappell• 12 years ago

I find it interesting that when the Google predictions match the CDC estimates poorly, the conclusion must be that Google got it wrong. But the CDC estimates are based on patient data that has been filtered through doctors' judgements, and those doctors only got to see the people that made use of health-care services. I think it is a quite reasonable possibility that Google's predictions were on the mark, and the CDC estimates were where the problem was.

Of course, this is a science article for the general population. I imagine there is significant analysis that were not being told about. Perhaps there are very good reasons to believe that it is Google, and not the CDC, whose estimates fail to match reality.

Still, I have to wonder.

Google Retools Its Flu Prediction Engine After Getting It Wrong

Related Stories