Bill Schmarzo envisions a future where holiday photos posted on Facebook become a gauge of a person’s weight loss or gain over time.
The chief technology officer for EMC’s consultant services acknowledges that privacy advocates are unlikely to allow his fantasy to become reality, but the technology that can measure minute body changes in photographs and feed it into someone’s electronic health record already exists.
“The scary thing is if that data could be used to deny care and insurability,” Mr Schmarzo says.
Healthcare companies are loathe to tread into such sensitive territory, but they are keenly aware of the gold mine of health data stored in people’s social media accounts.
“Studies have shown that people are more willing to share more private medical information in social media than they’re willing to share with their medical providers,” says Martin Kohn, chief medical scientist at IBM.
Pharmaceutical companies already analyse social media sites to track reports of side-effects of their drugs. That data can help correct formulas more quickly than waiting for the results of years-long clinical trials; it can also be used to set prices and test marketing slogans.
Public health officials are also interested in social media data as a source of information on disease outbreaks. Several start-ups are working on algorithms that study Facebook, Twitter, and blog posts to track early signs of infectious disease outbreaks, as people generally complain to their friends long before public health agencies can collect doctors’ reports and issue official warnings.
One outbreak of the Norovirus stomach bug at a student journalism conference in Canada was live-tweeted earlier this year, with posts such as “Motion that nobody else on this bus puke” and “36 hours without leaving my hotel room” signalling its coming and going before any traditional surveillance system had noted it.
The trouble with social media data are that they are fragmented and incomplete, warns James Kaufman, manager of public health research at IBM’s Silicon Valley lab, and there are limits to the types of insights that can be drawn from such patient-reported data.
“Colds, flus, sure people report that,” he says, “but no one’s going to report Aids or haemorrhoids.”