Twitter is often been hypothesised as a tool which may help to track disease. As I’m sure everyone is now aware, over the last week it has become clear that a flu outbreak has occured in Mexico, whether this new strain of flu turns into a pandemic is yet to be established. With the news crystalising over the weekend it was encouraging to see the unrelenting productivity of the internet at work. People quickly began to make maps visualising cases of the outbreak.
They’re great for tracking news stories visually but is it possible that as interesting and useful as these methods are, they’re now outdated? Traditional news outlets have been proven to be behind the curve in tracking major news events since twitter’s defining moment when it broke the news about the Mumbai terrorist attacks.
I expect we’ll have a map mashup of “flu” like tweets any time now but one programmer already has an interesting application in place to provide a similar service.
SickCity is a bottom-up, realtime alert system for outbreaks of infectious disease in your city, configurable by anyone and costing essentially nothing to run.
SickCity monitors twitter and facebook updates for terms like “flu”, “chicken pox”, “fever” within specific cities then plots those results over a 30 day graph so you can get an idea of how your city is doing in health terms. Looking at SickCity this morning suggests there has been a rise in symptoms in New York, an state which now has confirmed cases of the outbreak, whilst London, a city with no confirmed or suspected cases, shows only an upwards trend in people talking swine flu.
It provides links to recent tweets suggest people are falling ill, but is the service hyperbole or actually useful? In February Google published research suggesting they could track flu outbreaks quicker than the US Center for Disease Control and Prevention could based on what people were searching for. It was compelling enough to interest Nature, the leading scientific journal, who published the findings in February. Tracking flu pandemics based on what people are searching for is a tried and tested method, so there’s no reason to believe that tracking what people are tweeting should be any less relevant.
There are any number of ways it can be done, you could set up a simple search term in your tweetdeck to monitor what people are saying but how to filter out the noise of people simply commenting about the story? An advanced twitter search for terms “sore throat” “ill” “fever” exclulding the term “swine” and within 500 miles of London produces these results. Hard to analyse what is happening when all you’re left with is raw data, twitter only gets really interesting when it’s data is aggregated in an interesting way. Sites like Twist don’t like advanced search queries and are only useful for simple search terms, so I’d like to throw down a challenge, what can you come up with to visualise these twitter search terms? Twitter need to have a system that visualises advanced search queries by 9AM yesterday, but currently the advanced search produces only raw data.
As Hans Rosling says, all of the data is there; it’s being tapped out by millions of tweeters all over the globe.
Like this post? Tweet it with our Bit.ly link and you’ll show up in our conversation tracker!