Google flu tricked: Any academic research about this topic?

Hello guys,

several articles came out about a short wile ago:

http://time.com/23782/google-flu-trends-big-data-problems/

http://www.businessweek.com/articles/2014-04-17/the-flu-tricked-google-dot-can-wikipedia-do-better

http://www.nature.com/news/when-google-got-flu-wrong-1.12413

As we can see from this example, google flu tends and also a lot of other models tend to have the problem of losing their accuracy over time, the model tends to fit to good to the underlying data(curve-fitting). Is there any scientific research done in this field how to prevent this phenomenon?

Would appreciate papers/keywords!