Artificial Intelligence Could Help Prevent Suicide, Says New Study

Scientists have created a machine-learning algorithm that could help prevent suicides. In a study published in Nature Human Behavior, researchers at Carnegie Mellon and the University of Pittsburgh studied the brain patterns of suicidal individuals on a fMRI machine and examined how they thought about life and death. They then developed an algorithm that could track these signals. For example, the algorithm learned to isolate frontal lobe flares at the mention of the word death.

The program was soon able to pinpoint suicidal markers with a high degree of accuracy, over 90 percent accuracy to be more precise.

But as Wired notes, the study was not without its drawbacks. There was a small sample size, only 34 people, so it’s still unclear whether the algorithm would work when applied to a larger population. Also, fMRI studies tend to have certain inherent flaws. One of them is related to questions around causation theory i.e. just because two things happen at the same time, doesn’t mean that one caused the other. So, just because a frontal lobe flared at the mention of the word death doesn’t mean that the word caused the flare.

Furthermore, given the complexity of the brain, there are issues around scientists deciding that a part of the brain does something and then confirming their hypothesis linked to a “hand-picked set of triggers.”


For the study, researchers recruited 17 adults between the ages of 18 and 30 who had recently reported suicidal thoughts to their therapists. Then they picked 17 neurotypical participants for the control and placed each group inside of a fMRI scanner. During the scan, each test subject was shown a random series of 30 words. These were a mixture of positive words, negative words, and words linked to death and suicide. The subjects were then asked to think about the word for three seconds, with scientists recording the blood flow to the brain to find out which parts of it were working when the word was mentioned.

They then took the data from the brain scans and loaded them into a machine-learning program. With each word, they told the algorithm which of the results came from the test group (suicidal thoughts) and which ones came from the control group. But they also left out one person at random. They then trained the program to tell the two groups apart based on their brain signals.

According to Wired, when key topics were examined like death, cruelty, trouble, carefree, good, and praise, the program was able to tell whether the left out person was a suicidal ideator or not.


Do you think that this algorithm will reduce the high number of suicides that happen in the U.S.? Let us know what you think in the comments below.

[Featured Image by Tatiana Shepeleva/Getty Images]