This AI can spot suicidal thoughts and signs of depression from a person’s brain

In what has the potential to be a life-saving breakthrough, AI has been used to identify people with suicidal thoughts from their brain scans.

This AI can spot suicidal thoughts and signs of depression from a person’s brain

Using a neural decoder previously trained to identify emotions, as well as complex thoughts, researchers from the University of Pittsburgh and Carnegie Mellon University have developed an algorithm that can spot signs of suicidal ideation and behaviour.

The study was led by Marcel Just and David Brent. The pair devised a list of 30 words – 10 “death-related” words, including the word death itself, 10 positive words such as “carefree” and “praise”, and 10 negative words including ”trouble” and “cruelty”. These words were then read to 17 people with known suicidal tendencies, and 17 so-called “neurotypical” people – people with no known history of related mental health issues, depression or suicidal thoughts.

The researchers applied their machine-learning algorithm to brain scans taken from these participants and the software correctly identified which group the participants were from with 91% accuracy using changes in their brain activation patterns.

READ NEXT: Spike in suicide searches linked to 13 Reasons Why

A follow-up test saw the AI being trained specifically on the brain scans of those in the group linked with suicidal thoughts to see if the software could identify those who had previously attempted suicide. It was correct in 94% of cases.


Going a step further, to understand what caused these differing brain patterns, Just and Brent linked the patterns to previously gathered neural signatures used to identify emotions, particularly sadness, shame, anger and pride. The AI was able to measure “the amount of each emotion” that each of the words generated in the brains of participants.

During this test, the machine-learning algorithm was less successful but still impressive. It was able to accurately predict which group the participant belonged to with 85% accuracy using just the differences in the emotion neural signatures.
There are limitations with the research, of course. It was only trialled on a small sample group and the team said it now plans to test the approach on a larger group. This will help determine the software’s “generality and ability to predict future suicidal behaviour.”

READ NEXT: Suicide, depression and the tech industry

Brent hopes that the results of these more widespread trials could even help doctors identify, monitor and intervene when people are at risk in the future.

“Our latest work…gives us a window into the brain and mind, shedding light on how suicidal individuals think about suicide and emotion-related concepts. What is central to this study is that we can tell whether someone is considering suicide by the way that they are thinking about the death-related topics,” said Just.

If you or a loved one has been affected by the issues raised in this story, you can get support and advice over in our online help and support guide. You can also reach the Samaritans for free 24 hours a day on 1116 123 or the support group Campaign Against Living Miserably (CALM) specifically for young men at 0800 585858.

Disclaimer: Some pages on this site may include an affiliate link. This does not effect our editorial in any way.

Todays Highlights
How to See Google Search History
how to download photos from google photos