Child and Adolescent Self-Destructive Thoughts and Behavior: Conceptualizing and Holistic Risk Assessment Using Transparent Artificial Intelligence

Although adolescent suicidal thoughts and behavior is a pressing global concern, conceptual frameworks for understanding suicidality in youth remain limited. The theory of adolescent suicidality and self-destruction (Haghish, 2024) identifies six risk and protective domains for suicidal and self-destructive behavior. However, several of the hypothesis of this data-driven theory have not been yet assessed. Using already collected datasets and holistic transparent AI algorithms such as HMDA (Haghish, 2025c), this proposal aims to further test the theory across new populations and age groups. Five research questions guide the work, focusing on predictors of passive self-destructive behavior, distinctions between passive and active self-destruction, validation of the theory in children, and psychometric assessment of fairness in AI risk modeling. Four already collected datasets from public domain and R-BUP , and the UiB supervisor will be used. AI analysis will be used for holistic domain analysis and differential diagnosis. This interdisciplinary research holds substantial potential to enhance theoretical understanding and inform more effective, targeted suicide prevention and intervention efforts for young populations. It also develops an innovative solution and algorithm to address the utility and ethical concerns regarding AI decision-making in health psychology, contributing to both artificial intelligence and psychology.