Artificial Intelligence May Be Able To Detect Suicidal Tendencies From Brain Scans

The clinical assessment of suicidal tendency is currently done only by a psychological assessment which essentially involves asking the question — Are you suicidal?

According to a clinical research published by United States National Library of Medicine in a study called “Clinical Correlates of inpatient suicide,” four in five patients who died by suicide had denied such tendencies in psychological assessment.

According to a new study titled “Machine learning of neural representations of suicide and emotion concepts identifies suicidal youth” by researchers from the University of Pittsburgh and Carnegie Mellon University, artificial intelligence models may succeed where psychologists don’t in detect such tendencies.

The researchers put 17 adults with suicidal ideation using 17 control subjects who were not known to have such tendencies and put them in fMRI scanner to measure which areas of the brain would be activated when the subjects would be thinking about those keywords.

For example, the study states that words like death and cruelty activated left the superior medial frontal area of the brain and the medial frontal/anterior cingulate respectively.

The researchers then analyzed the results using a machine learning algorithm and found that they could map suicidal tendencies more accurately than standard risk assessments.

"Our latest work is unique insofar as it identifies concept alterations that are associated with suicidal ideation and behavior, using machine-learning algorithms to assess the neural representation of specific concepts related to suicide. This gives us a window into the brain and mind, shedding light on how suicidal individuals think about suicide and emotion-related concepts,” Marcel Adam Just, the lead author on the study and a professor of psychology at the Carnegie Mellon University stated in the press release.

"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," he further added.

The researchers claim that their method, on further testing, was able to detect suicidal tendencies with 94 percent accuracy. The program run using the machine learning algorithm could easily distinguish between subjects who had suicidal tendencies and ones who didn’t.

This method could be very helpful in psychological evaluations, according to researcher David Brent.

“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," he stated in the press release.

However, the model can only gage patients’responses to stimuli — it is not foolproof. If patients want to conceal information or give wrong responses, they can still do that, which would affect the accuracy of the algorithm. But it is expected to map brain activity properly and provide an accuracy in detection in many cases that could have been surpassed by clinical psychologists.