What if it was possible to predict which people are at high risk of suicide? Researchers may have brought us a step closer to such a feat, after developing a brain imaging technique that could pinpoint individuals with suicidal tendencies.
Risk factors for suicidal behavior include feelings of depression, anxiety, and stress, a history of mental illness, and a history of drug or alcohol abuse.
Demonstrating increased aggression, isolation, or a greater use of alcohol or drugs, or talking about suicide or being a burden to others can be warning signs for suicide.
However, the only way to truly know whether a person will attempt suicide is to pinpoint what is happening in their minds. A new study may have uncovered a way to do just that.
Researchers from Carnegie Mellon University (CMU) and the University of Pittsburgh have created a brain imaging method that can accurately distinguish between individuals with and without suicidal thoughts.
Using fMRI to predict suicide risk
For their study, the researchers enrolled 34 participants. Of these, 17 had suicidal tendencies and 17 were control subjects.
The participants were all presented with three lists of 10 words. One included words with negative associations (such as “evil,” “cruelty,” and “trouble”), one included positive words (such as “good,” “carefree,” and “praise”), while the third included words related to suicide (such as “death,” “hopeless,” and “distressed”).
As the subjects were shown the word lists, they underwent functional MRI of the brain, which enabled the researchers to monitor their neural response to each word.
The scientists found that the subjects’ neural response to six words — “death,” “cruelty,” “trouble,” “carefree,” “good,” and “praise” — across five specific brain regions were best for distinguishing between participants with suicidal tendencies and the controls.
By training a “machine-learning algorithm” to use these data, the researchers found that they were able to identify subjects with and without suicidal tendencies with 91 percent accuracy.
Next, the team divided those with suicidal tendencies into two groups: those who had attempted suicide and those who had not. They found that their algorithm was able to distinguish between these two groups with 94 percent accuracy.
Identifying the emotions behind the words
The researchers then set out to determine the mechanisms behind the varying neural responses between participants with suicidal tendencies and the control group.
Specifically, they wanted to find out what emotions were evoked when subjects thought of the six words that were used to pinpoint suicidal ideation and behavior.
To reach their findings, the team added neural signatures for different emotions — including sadness, anger, shame, and pride — to their machine-learning algorithm.
They found that the new algorithm was 85 percent accurate in identifying which subjects had suicidal tendencies.
“The benefit of this latter approach,” says Just, “sometimes called explainable artificial intelligence, is more revealing of what discriminates the two groups, namely the types of emotions that the discriminating words evoke.”
“People with suicidal thoughts experience different emotions when they think about some of the test concepts,” he continues. “For example, the concept of ‘death’ evoked more shame and more sadness in the group that thought about suicide. This extra bit of understanding may suggest an avenue to treatment that attempts to change the emotional response to certain concepts.”
A tool for predicting suicide?
The researchers note that their results need to be replicated in a larger cohort, but they believe that the technique holds promise for pinpointing individuals who are at high risk of suicidal behavior.
“Further testing of this approach in a larger sample will determine its generality and its ability to predict future suicidal behavior, 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.” – Study co-author David Brent, University of Pittsburgh
Barry Horwitz — the chief of the Brain Imaging and Modeling Section at the National Institute on Deafness and Other Communication Disorders comments on the algorithm in an editorial accompanying the study. He says that if the study results are confirmed in future research, “then a case can be made that functional neuroimaging has potential to become a major medical tool for diagnosis and/or evaluation of treatment efficacy of psychiatric disorders.”