Detecting depression and mental illness on social media: an integrative review. (December 2017)

• Record Type: Journal Article
• Authors: Guntuku, Sharath Chandra; Yaden, David B; Kern, Margaret L; Ungar, Lyle H; Eichstaedt, Johannes C
• Is Part Of: Current opinion in behavioral sciences. Volume 18(2017)
• Page Start: 43
• Page End: 49
• Publication Date: 2017-12
• Subjects: Psychology -- Periodicals; 150.5
• DOI: 10.1016/j.cobeha.2017.07.005
• Languages: English
• ISSNs: 2352-1546

Abstract

Highlights: Mental illness is underdiagnosed but observable in online contexts. Characteristic language use patterns associated with depression have been identified and allow for the detection of mental illness with mixed performance. Prediction accuracies fall between unaided clinician assessment and screening surveys. No studies to date are based on gold-standard clinical diagnoses. The findings are still preliminary and the field relatively nascent. Abstract : Although rates of diagnosing mental illness have improved over the past few decades, many cases remain undetected. Symptoms associated with mental illness are observable on Twitter, Facebook, and web forums, and automated methods are increasingly able to detect depression and other mental illnesses. In this paper, recent studies that aimed to predict mental illness using social media are reviewed. Mentally ill users have been identified using screening surveys, their public sharing of a diagnosis on Twitter, or by their membership in an online forum, and they were distinguishable from control users by patterns in their language and online activity. Automated detection methods may help to identify depressed or otherwise at-risk individuals through the large-scale passive monitoring of social media, and in the future may complement existing screening procedures.