New computer technology can help identify someone as suicidal based on verbal and
nonverbal cues, researchers found.
“These computational approaches provide novel opportunities to apply technological
innovations in suicide care and prevention, and it surely is needed,” lead author
John Pestian, Ph.D., said in a news release. “When you look around health care facilities,
you see tremendous support from technology, but not so much for those who care for
Researchers recruited adolescents and adults from emergency departments and inpatient
and outpatient centers to participate in a prospective clinical trial, and 371 completed
the study. Participants were suicidal, nonsuicidal with a mental illness or controls.
The subjects filled out the Columbia-Suicide Severity Rating Scale, Young Mania Rating
Scale and Hamilton Rating Scale for Depression. They also took part in a recorded
interview, answering open-ended questions about whether they had hope, fear, secrets,
anger or emotional pain.
A computer algorithm then analyzed linguistics and acoustics from the interview and
was able to categorize someone as suicidal with 93% accuracy. The computer also could
correctly characterize someone as suicidal, mentally ill but not suicidal or neither
85% of the time.
Researchers noted that the mentally ill and control patients tended to laugh more
during interviews, sigh less, and express less anger, less emotional pain and more
hope than the suicidal patients.
“Overall, the results show that machine learning algorithms can be trained to automatically
identify the suicidal subjects in a group of suicidal, mentally ill, and control subjects,”
they wrote. “Moreover, the inclusion of acoustic characteristics is most helpful when
classifying between suicidal and mentally ill subjects.”
In addition to the health care setting, such technology could be useful in schools,
youth clubs and juvenile justice centers.
“These computational approaches may provide novel opportunities for large-scale innovations
in suicidal care,” authors wrote.