A team from the Universities Intelligent Systems Laboratory have built a predictive model that was able to identify keywords in tweets associated with flu “infections” and estimate the severity of the disease in a specific region or area.
The researchers agree that Twitter user do not represent the population in general but it could still be a very valuable tool in tracking flu related events.
Professor Nello Cristianini, who carried out the research said “Our research has demonstrated a method, by using the content of Twitter, to track an event when it occurs and the scale of it”
“We were able to turn geo-tagged user posts on the micro blogging service of Twitter to topic-specific geo-located signals by selecting textual features that showed the content and understanding of the text.’
Over several months, the team of researchers gathered a database of more than 50 million geographical located “tweets” which could then be compared to official data from the NHS on flu incidence by location / region.
A state-of-the art machine-learning algorithm automatically figured out which keywords in the database of tweets were associated with elevated levels of flu.