Thank you for using OutSmart Flu. Your reports are helping us track and report the spread of flu on campus.
The first of three $500 raffles was held on October 31. The winner was a UW senior majoring in microbiology. Congratulations!
The next raffle will be held on December 20. Good luck, OutSmart Flu users!
So far, we have about 500 OutSmart Flu users. Thank you for being citizen scientists and helping to track flu on campus.
Flu activity continues to be low in much of the U.S. The continued growth of our community of users will improve the accuracy of estimating flu on our campus. Please invite your friends to join this unique initiative.
Google recently revised their algorithm to estimate flu activity in the U.S. and retrospectively revised their Flu Trends data.
Check out the Flu Trends graphs below showing data from before and after revision of the algorithm. Pay attention to the bold blue line.
Oct 24 estimates of flu activity this year (bold) vs. previous years (thin), Madison, WI
Nov 1 estimates of flu activity this year (bold) vs. previous years (thin), Madison, WI
First, some background: The Google method is to analyze how many people are googling keywords like “flu” or “influenza.” In 2010, they showed that this method can predict flu epidemics in the U.S. 1-2 weeks before methods used by the CDC. Last year, flu transmission reached intense levels in December-January. During this time, the Google method overestimated the incidence of flu in the U.S. by more than double. Google’s estimates of flu activity during that period were still timely, but not accurate.
What went wrong?
The Google algorithm is susceptible to heightened media coverage of flu. When flu transmission is intense, media covers it, and more people google flu-related topics even if they are feeling well. Google knew of the vulnerability before the 2012-13 flu season. After its effect was observed, they could revise and try a new algorithm, and they did so last week.
Opportunity for Citizen Science
Limitations in Google’s method are why we began OutSmart Flu. Analyzing symptom reports is a more accurate method to estimate flu activity than mining search engine data.<br><br>Our goal is to provide you timely and accurate information on flu spreading at UW. You are helping UW become the first to crowd-source campus flu activity using smartphones. This can open the door to crowd-sourcing the detection of other illnesses on campus.
Please continue to send us feedback. You are helping us to improve the app and the OutSmart Flu initiative at UW-Madison!