Researchers have developed a new machine with a learning-based approach to recognise adolescents undergoing suicidal thoughts and behaviour
Suicidal thoughts and behaviour, often recognised through factors including online and at-school bullying, serious arguments at home, gender, alcohol use and attitudes about marijuana, are greatly influential during adolescence.
The identification of specific risk factors associated with suicidal thoughts and behaviour among adolescents helps to improve suicide prevention efforts.
Few studies have ever investigated these risk factors in combination with each other, especially in larger cohorts of adolescents. Therefore, this specific kind of machine learning establishes new opportunities for research, which could ultimately advance prevention efforts.
“Strong predictive accuracy” says research team
Researchers applied a machine-learning analysis to data from a survey of high school students in Utah, who routinely conduct student monitoring for issues such as drug abuse and mental health.
The researchers discovered that they could predict 91% accuracy using the data of individual adolescents’ answers, indicating suicidal thoughts or behaviour.
The data included responses to more than 300 questions each for more than 179,000 high school students who took the survey between 2011 to 2017, as well as demographic data from the U.S. census.
The survey questions which had the most predictive ability were questions about digital media harassment or threats, at-school bullying, serious arguments at home, gender, alcohol use, feelings of safety at school, age, and attitudes about marijuana.
With the new algorithm’s accuracy being higher than any previously developed predictive approaches, researchers suggested that machine-learning could indeed improve understanding of adolescent suicidal thoughts and behaviour.
The authors said: “Our paper examines machine learning approaches applied to a large dataset of adolescent questionnaires, in order to predict suicidal thoughts and behaviours from their answers. We find strong predictive accuracy in identifying those at risk and analyse our model with recent advances in ML interpretability.
“We find strong predictive accuracy in identifying those at risk”
“We found that factors that strongly influence the model include bullying and harassment, as expected, but also aspects of their family life, such as being in a family with yelling and/or serious arguments. We hope that this study can provide insight to inform early prevention efforts.”
Orion Weller of Johns Hopkins University in Baltimore, Maryland, and colleagues presented their results in the open-access journal PLOS ONE on November 3rd, 2021. Future research in this field could develop the study by using data from other states in the U.S, as well as data on actual suicide rates.
