Manchester is known as the heart disease capital of Britain. Tamworth has been dubbed the worst place for obesity. New research, however, suggests there is a better tool for tracking health issues than geography: Facebook.
Profiles on the social network could be ten times more effective at predicting health problems such as from smoking or excess weight than looking at where people live, according to data scientists.
Alex Kogan, a social psychologist, and his colleagues at the University of Cambridge have built an algorithm that can “learn” Facebook users’ political preferences, health risks and levels of happiness from their “likes”, interactions and friendship groups.
So far they have found strong links between users’ Facebook friends and their tendencies to smoke and drink, and have diabetes, high cholesterol and high blood pressure.
Most of the work is done by a computer that is fed large-scale survey responses from the US and then crunches people’s profiles to work out, for example, whether retweeting the Pope’s Latin Twitter account makes you more or less likely to vote for Mitt Romney or liking “OMG These Monkey Videos Will Make You Weep!!!” on Facebook suggests you are a smoker.
Advertisement
Dr Kogan will present his early findings at the Cambridge Science Festival.