“Tweets” are the short messages that users of Twitter share with friends and the general public. In April 2009, the authors of this study began collecting and storing public tweets related to H1N1, or swine flu. The tweets matched a set of general keywords (e.g., flu, swine, tamiflu); additional keywords located tweets that mentioned travel and hygiene. Then, in October 2009, based on discussions with public health officials, the authors expanded their search terms to follow concerns about vaccines.
During the study, a digital map of the U.S. continuously displayed the time and location of the most recent 500 tweets. By passing their computer cursor over tweet, viewers of the map could instantly read messages related to H1N1 (article includes a sample map image).
The costs of tracking disease outbreaks in real-time based on emergency department data are prohibitive. This study established a model for using Twitter to monitor disease activity.