Artificial intelligence could one day be used to help identify a person contemplating suicide. Around 800,000 people die a year from suicide, according to the World Health Organization. It’s currently the second-leading cause of death in the U.S. among people between the ages of 15 and 24.

But a new study is using brain scans and AI to show how someone experiencing suicidal thoughts thinks differently about life and death. The results were published in Nature Human Behavior.

“What is central to this new study is that we can tell whether someone is considering suicide by the way that they are thinking about the death-related topics,” Marcel Just, the study’s co-lead author, told CMU.edu.

Researchers looked at 34 participants, 17 of whom were experiencing suicidal thoughts. Each person underwent brain imaging with an MRI machine while being shown a series of words that were related to different emotions. Ten words had a positive emotional relation, such as “carefree,” 10 were negative, like “trouble,” and 10 were related to suicide, including “desperate” and “hopeless.”

Based on the participants that were already considered suicidal, the researchers were able to map five regions of the brain and six words that could identify someone as suicidal. Next they devised an algorithm to spot the same thing.

The algorithm correctly identified 15 of the 17 suicidal patients and 16 of the 17 members in the control group.

Researchers then applied the algorithm to the group of participants experiencing suicidal thoughts – half had previously attempted suicide and half had not. The algorithm correctly identified 94 percent of the participants who had attempted suicide.

“Further testing of this approach in a larger sample will determine its ability to predict future suicidal behavior,” said David Brent, co-head author of the study. “And could give clinicians in the future a way to identify, monitor and perhaps intervene with the altered and often distorted thinking that so often characterizes seriously suicidal individuals.”

Article By Lauren Tousignant