Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Predicting Suicidal Ideation Among U.S. Adolescents Using Supervised Machine Learning and AI-Driven Deep Learning
ABSTRACT
Background:
Youth suicide remains a critical public health crisis in the United States. We aimed to develop predictive models to identify risk factors for suicidal ideation using supervised machine learning techniques, which could aid in early intervention efforts and help prevent suicide. We also tested an AI-driven deep learning model for potential better performance.
Objective:
The objective of this study was to predict factors associated with suicidal ideation using predictive models from sociodemographic factors and selected risky health behaviors and adverse experiences among U.S high school students. Additionally, the study compared the predictive performance of an Artificial Neural Networks model with supervised machine learning models.
Methods:
We analyzed data on 7,612 U.S high school students, including 1,598 (20.99%) reported suicidal ideation, from the 2023 Youth Risk Behavioral Surveillance System. We developed ten machine learning models including Decision Tree, Naïve Bayes, Random Forest, Logistic Regression, Support Vector Machine, Extreme Gradient Boosting, Neural Network, and Artificial Neural Network. The optimal model was selected based on accuracy, sensitivity, specificity, receiver-operating characteristics area under the curves (ROC AUC), precision, and F-1 score. Feature importance scores were used to identify the critical predictors of suicidal ideation.
Results:
All supervised machine learning models achieved high accuracy (82.4%-84.9%) and high AUC (84.6%-88.0%). Random Forest had the highest accuracy (84.9%) and AUC (88.0%). Artificial Neural Network resulted in slightly better accuracy (85.2%). Overall, Random Forest was the optimal model. Key predictors of suicidal ideation included feeling consistently sad or hopeless, poor mental health, being bullied, violence-related experiences and school racism. Risky health behaviors and substance use were moderately important risk factors. School connectedness showed a protective effect.
Conclusions:
Supervised machine learning was a promising tool for risk assessment and early intervention. AI-driven deep learning showed potential for improving predictive accuracy, though further validation studies are needed for use in public health models.
Citation
The author of this paper has made a PDF available, but requires the user to login, or create an account.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.