A study by Inserm, unveiled on Tuesday and published in
, looked at 70 predictors of suicidal behavior in students.
Thanks to questionnaires completed one year apart, artificial intelligence has made it possible to list four fundamental factors.
These results, which remain to be completed, could be used in the future to create questionnaires, more restricted and less intrusive, to identify and best support students with psychological difficulties.
It is the second leading cause of death among 15-24 year olds.
Suicide remains a taboo subject, and yet, better support and better prevention of these acts among young French people is urgent.
And to do this, researchers could provide part of the solution.
A study, which has just been published in
opens up new perspectives.
Both for research and for the prevention of suicide among young people.
And this thanks to artificial intelligence.
Four factors to predict the risk of suicidal thoughts in students
A team of researchers from Inserm and the University of Bordeaux, in collaboration with the universities of Montreal and McGill in Quebec, has identified four mental health indicators that accurately predict suicidal behavior in students. How was this study constructed? From a cohort, followed between 2013 and 2019 throughout France, comprising 20,000 students with an average age of 21 years. Among them, 5,000 responded to two questionnaires, a year apart. And about 17% of participants exhibited suicidal behaviors in the year between these two questionnaires. Without difference between the sexes: 17.4% for girls, 16.8% for boys.
“This is not trivial, it's almost 1 in 5, underlines Mélissa Macalli
doctoral student in epidemiology at the University of Bordeaux and co-author of the study.
But this corresponds to the prevalences found in other studies.
And the risk of suicide is known to be high among college students.
The present, not the past
To obtain the most precise results possible, the students answered 130 questions: age, sex, sectors, sports activities, housing, addictions, childhood, family history, mental health
“Among 70 factors, we looked at which were the best. predictors of future suicidal behavior.
“Without having a number in mind.
Surprise: Four of these factors make it possible to detect approximately 80% of suicidal behaviors during follow-up: suicidal thoughts, anxiety, symptoms of depression and self-esteem.
“I have worked a lot on mental health and childhood;
I expected to see strong predictors of mistreatment, parental support, insists the researcher.
Which are known factors in suicide.
Now, the four major factors have to do with the present, not the past.
“In boys, self-esteem is almost the only predictor, which is a bit unexpected,” she continues.
Build a tool to predict suicide risk
This study would not have been possible without the help of artificial intelligence. Because with 130 questions and 10,000 questionnaires, the analysis of these data required a titanic work… “The traditional statistical models do not support so many variables, specifies Mélissa Macalli. Moreover, they do not give precise predictions on the question of suicide. This is the first study with AI in mental health for students. "
What is the use of such research?
Ultimately, these results, provided they are confirmed and supplemented by a study on other students, could feed a prediction tool.
“Currently, nothing is done to detect suicidal students, regrets the doctoral student.
We can imagine in the future offering a short online questionnaire, which would identify those at high risk.
»And in this case, offer them support.
Support from "bridging students"
“Suicidal behavior is a very complex multifactorial process,” she continues. You can't ask thousands of intrusive questions of every returning student. That is why we were trying to find out what would make it possible to detect them without having to ask them if they had been mistreated. ”It is not easy, however, to put a student in a“ suicide alert ”box on the pretext that she has poor self-esteem (and this is common…). “It is a very sensitive subject: we must both reassure them and alert them, recognizes Mélissa Macalli. This is why we must test our tool. It is not a question of detecting false positives. Nor to say "beware, this test will tell you if you are going to commit suicide". "
To be as close as possible to the concerns of these young people, the team worked with psychiatrists, but also “relay students”.
“They participate in the construction of the tools.
At 42, I have neither their language nor their mentality.
On the other hand, we try to fight against the taboo, the stigmatization.
We realized that inviting people to talk about suicide relieves them.
We insist: it can happen… and it can stop!
And every suicidal thought does not lead to acting out.
A complementary study on young people during the epidemic
The health crisis has put a spotlight on the importance of taking care of the mental health of young people. A subject that Mélissa Macali knows well. Since March 2020, it has carried out a complementary study. This time on the mental health of young people, with questionnaires every month. Out of a cohort of 4,000 people, aged 18 to 40, half of the respondents study, the others work. “It has been observed that students have significantly more degraded mental health scores. They are twice as likely to present symptoms of anxiety and depression as workers: 33% against 16%. With a specific increased risk during confinement. "
In the summer of 2020, the returns of the two populations were similar.
“But in October, the students sank,” she says.
From this summer, these monthly reports will be sent, via an application, to participating students.
“They were in demand during the epidemic to know where they were on mental health.
Coronavirus: Free shrink sessions for children, a useful device ... but insufficient
Schizophrenia: Informing young people to fight against prejudice and confusion