Facebook posts can predict who has diabetes

The social media platform is often better than demographic data at spotting illness: study

Trawling through Facebook posts could help identify which patients have diabetes, with US research showing the language used in posts can predict 21 medical conditions.


In fact, Facebook is better than demographic data (age, sex and race) at predicting 10 of the conditions, researchers from Penn Medicine at the University of Pennsylvania have found.

They report in PLOS One that the social media platform used by more two billion people is superior for flagging diabetes, with common red-flag words including “god”, “pray” and “family”.