Navigating the emotional maze: Understanding Adolescent suicidal ideation using CNN-LSTM model

Abstract

Teenage suicidal ideation is on the rise, which emphasizes how crucial it is to recognize and comprehend the variables that contribute to this problem. Convolutional neural networks (CNNs), which are complex machine learning models capable of analysing intricate relationships within a network, are one possible strategy for addressing this issue. In our study, we employed a CNN-LSTM hybrid model to explore the complex relationships between teen suicide ideation and various risk variables, including depression, anxiety, and social support by analysing a substantial dataset of mental health surveys, seeking patterns and risk factors associated with suicidal thoughts. Our objective was clear: identify adolescents prone to suicidal ideation. With 24 parameters and a sample size of 3075 subjects, our model achieved an impressive F1-score of 97.8%. These findings provide valuable insights which helps in developing effective preventive interventions to address adolescent suicidal ideation, finding out the important patterns and risk variables related to suicidal thoughts. The study results offer important direction for developing preventive interventions that successfully address adolescent suicidal ideation.