397 – Health Surveillance via Social Media
Quantifying Suicidal Ideation via Language Usage on Social Media
Glen Coppersmith
The Johns Hopkins University
Ryan Leary
Eric Whyne
Tony Wood
Suicide is a large and growing problem, yet relevant data to draw informed decisions and assess intervention strategies is sorely lacking, and often at least two years out of date. We analyze publicly available data to assess the viability of using it to provide more timely information. We examine quantifiable signals related to suicide attempts and suicidal ideation in the language of social media data. Our data consists of Twitter users who have attempted suicide and age- and gender-matched neurotypical controls and similarly matched clinically depressed users. We apply simple language modeling techniques to separate those users automatically, and examine what quantifiable signals allow them to function, tying them back to psychometrically validated concepts related to suicide. We then use these scalable classifiers with public social media data and open government data to suggest some direction for future epidemiological research. All this research is done with public data, though we take great care to protect the privacy of the users.