Got the flu? Google will find out long before your GP

It may be indelicate to kick off a column with a mention of urine, but the matter crops up repeatedly when people talk, blog or tweet about hospitals. And if a ward smells like a lavatory, we can conclude that matron isn’t being sufficiently fastidious about cleanliness.

All this is of enormous interest to Felix Greaves, an imaginative young doctor who has developed a computer algorithm to scour social media and spot “good” and “bad” hospitals from the tweets and Facebook posts they attract. Finding the word “urine” in the text is like hitting the jackpot – a near guarantee that the care has been more Dr Crippen than Dr Kildare.

Greaves’s study of “sentiment analysis”, buried away in the latest issue of the online journal BMJ Quality and Safety, is one example of the way that the online world is colliding with the offline worlds of science and medicine. Dr Greaves is exploring whether a rash of negative online comments can be used to flag up failing hospitals before they start killing patients.

Another example is Google Flu Trends: when people fall ill, they tend to type “flu remedy” or “influenza” into search engines. Because we go online in real time – i.e. you reach for your computer as the aches and fever begin to set in – Google Flu Trends can pinpoint outbreaks around one to two weeks ahead of traditional surveillance methods (such as totting up GP and hospital visits).

The US Department of Homeland Security is tracking social media for mentions of illness in order to predict public health crises; the website sickweather.com similarly sifts tweets and Facebook postings to create maps showing hotspots for colds, flu and even depression.

more here: Telegraph