Skip to main content
. 2021 May 17;16(5):e0250448. doi: 10.1371/journal.pone.0250448

Table 4. Example posts from a user(ui) ordered by timestamp (TS) and prediction from TvarM.

The italicized text are phrases which contributed to the representation of each post. These phrases had similarity to the concepts in medical knowledge bases.

Post 1 (TS 1): “Homie, … Im 27 yo, … the job is underpaying—700 euros per month … too afraid to search for a new job. … fuck me, I guess? … had these thoughts of suicide and these fears to take charge of my life from like the end of a high school. 10 years same feelings of dread, same thoughts of killing myself.”
Predicted Suicide Risk Severity: Suicide Ideation
Post 2 (TS 2): “One day …. sudden realization … I gonna gather determination … roll over the bridge. And my parents, or have a nice heart attack! feel trapped. … nothing gonna change. You will end up just like me. I will roll over the bridge
Predicted Suicide Risk Severity: Suicide Behavior
Post 3 (TS 3): “No wife, no house, no car, no decent job. Every single day … hating myself at work …. Im going to kill myself today or tomorrow. Probably … middle of next week, but the chances are … going to sleep forever
Predicted Suicide Risk Severity: Suicide Behavior
Post 4 (TS 4): “I dont even go to the exams … I might pass those exams … will not graduate …. playing some kind of a Illness jokemy poor family.”
Predicted Suicide Risk Severity: uninformative
User-level Predicted Suicide Risk Severity: Suicide Ideation