You are here
Jeremy Ginsberg, Matthew H. Mohebbi, Rajan S. Patel, Lynnette Brammer, Mark S. Smolinski, Larry Brilliant
The authors used an early version of google trends, which shows what people are searching for on google in a particular region; to measure where/when people had the flu. They then converted this data into a real time, region specific, data detection method, essentially attempting to see where and when people were searching for issues related to having the flu. Essentially, the authors were attempting to use google trends to predict where the flu was occurring. And they found a high correlation between people searching for flu-related terms, and actual physician visits related to having the flu. This model, the author’s claimed, successfully reported areas where the flu was prevalent in as little as one day, much sooner than the estimates provided by the CDC which averaged at least two weeks.
The policy implications for this type of work are intriguing. While this particular model has been criticized (see critique) for over predicting occurrences of the flu, for policymakers, the outcome is of small significance. What is of interest is the ability to predict attention relating to particular issues using google trends. Many politicians know how expensive public opinion polls are, and they are widely out of the reach of local elected officials due to price constraints, therefore, google trends could provide an interesting alternative for data savvy policymakers.
General/Not Specific Research
Find Similar Health Research