Could a brain scan predict suicide? Experts are cautious

New US study is one of many searching for signs of suicide risk in brain biology

Photograph: iStock

Photograph: iStock

 

American research into the prediction of the risk of suicide in at-risk individuals has provoked a cautious response from Irish experts in the field.

The study, published in Nature Human Behaviour, suggested calculating the risk of suicide could be substantially improved through the use of brain scanning techniques such as functional magnetic resonance imaging (fMRI).

It also suggested that machines with artificial intelligence could pick up the tell-tale neural signs of suicide risk through analysis of brain scans.

“There is optimism in our field that this approach can be useful in the future, but we’re not there yet,” said Prof Gary Donohoe, School of Psychology, NUI Galway, who is also part of the team at the Neuroimaging and Cognitive Genomics Centre at NUIG working on an international research project to better predict the course of diseases such as schizophrenia.

“In my opinion that will only happen when these kinds of analytical approaches can predict not just what diagnostic group you are currently in, but what group you are likely to belong to in the future,” said Prof Donohoe.

The researchers, who were led by Marcel Just, professor of psychology at Carnegie Mellon University in Pittsburgh, measured the response in people’s brains using fMRI to words with negative connotations, such as death and cruelty, as well as other words such as good and praise, which are considered positive.

The emotional reaction to such words was altered in the 17 people studied who had ideas of suicide, compared to 17 healthy individuals. “For example, death evokes more sadness and shame among suicidal ideators relative to healthy controls,” said Prof Just.

“Our technique has the capability of not only detecting that certain thoughts are altered in people with suicidal ideation, but it also provides information about the nature of the alteration. This information provides a target for treatment, which could aim to reduce or eliminate the alteration.”

Prof Just said his team used a machine learning mathematical algorithm (software designed to learn and get cleverer from experience) to identify people in the study with suicidal ideas based on their fMRI scans.

Risk assessment

The prediction of suicide risk for individuals is notoriously difficult, even for consultant psychiatrists with many years’ experience. There would be great interest, therefore, in any new technique that could reliably help clinicians to improve risk assessment in advance of discharging a patient, for example.

This latest study is part of a series of studies going on around the world where researchers are looking to measure the biological signs of mental ill-health and suicide risk in the brain biology of patients.

“This study does not add to the accuracy of risk assessment,” Brendan Kelly, professor of psychiatry at Trinity College Dublin, cautioned, “although it might point to a useful avenue for future work”.

The assessment of the risk of suicide for any individual in society over the course of their lifetime varies between 0.5 per cent and 1.2 per cent, said Prof Kelly.

For a person in Ireland today who lives to 90 years-old, the statistical risk of suicide is 0.93 per cent, he said, which corresponds to less than one in a hundred. That figure rises significantly to 10-15 per cent for individuals who have major depression.

Prof Kelly sounded a further note of caution saying that the sample size – the number of people that took part in the Carnegie Mellon study – was very small.

The only way to definitively show that using fMRI, or other brain scanning techniques to predict suicide risk worked, would be to scan people’s brains, make a risk prediction, and follow them up to see whether they did, in fact, die by suicide, said Prof Kelly. “That did not occur in this study,” he said.