Prediction and explanation of the increase in suicide risk of emerging adults: A comprehensive approach combining logistic regression, glasso network analysis, and Bayesian networks

BACKGROUND: The high suicide rate among emerging adults is a pressing public health issue. Identifying key suicide risk factors and understanding the mechanisms through which they affect individuals is crucial for intervention. AIM: To uncover the complex factors influencing suicide risk among emerging adults and to elucidate how these risk factors interact and contribute to the overall risk of suicide. METHODS: An online survey assessed mental health and suicide risk factors among 29,111 college students. Higher-risk students (n = 4820) were further evaluated using the Adolescent Suicidal Tendency Scale. This two-phase approach identified initial risk factors and subsequent suicide risk, which were analyzed through logistic regression, Glasso, and Bayesian network methods. RESULTS: Logistic regression results indicated that adverse life events and social support can predict suicide risk, with the model achieving an Area Under the Curve (AUC) of 0.783. Glasso network analysis revealed a highly interconnected symptom network among all factors, where the highest centrality nodes, such as depression (1.465) and neuroticism personality traits (1.139), played central roles in the evolving dynamics of suicide risk. The Bayesian network analysis emphasized the mediating role of social support in the relationship between other risk factors and suicidal ideation. LIMITATIONS: The lack of repeated measurements and the exclusion of pandemic-related variables may limit a comprehensive understanding of the risk factors. CONCLUSIONS: Intervening in the mental health issues of individuals with suicidal tendencies and strengthening social support are crucial for reducing suicide risk, and this deserves the attention of mental health professionals.