In a country where millions of Americans are afflicted with depression and PTSD, and where many of them do not get treatment (whether by intention, inability, or system failure), social media may become the key to diagnosing large populations who are suffering in silence, of a sort.
At the American Statistical Association national conference in Boston last week, a presentation titled “Depression, Economics, and Population in Social Media,” was given by Elizabeth Hohman, David Marchette, and Glen Coppersmith of the Navy and Johns Hopkins University. In it, they reviewed the Twitter traffic of people across the country, including the posts of those suffering from PTSD and Seasonal Affective Disorder (SAD) and how they differed from those of the general tweeting population. The researchers looked at word usage and applied a “depression classifier” algorithm to the content of millions of brief messages. They found that those living in areas of high unemployment produced Twitter content correlating with depression, and that PTSD-related content could be traced to areas with large populations of veterans.
Carried forward and further refined, such work could revolutionize large-scale public-health efforts in behavioral healthcare, so that it will be possible to find untreated victims by listening for 140-character cries for help.