AI-Based Psychodiagnostics’ Models to Support Early Intervention and Reduce Suicide Risk in Adolescents and Youth: Development and Clinical Validation

Abstract

Adolescent suicide has become a major public health concern worldwide, requiring more effective approaches for early detection and prevention. This study examined the development and validation of artificial intelligence (AI)–based psychodiagnostics models designed to support early identification of suicide risk among adolescents and youth. A quantitative research design was employed using structured psychodiagnostics assessment data collected from 268 adolescents aged between 13 and 21 years who had undergone psychological screening in educational or mental health support environments. The study analyzed key psychological and behavioral constructs including depressive symptoms, emotional dysregulation, impulsivity, and social stress exposure to determine their predictive influence on suicide risk indicators. Descriptive statistical analysis revealed moderate levels of psychological distress within the study population, with depressive symptoms recording the highest mean score (M = 3.45, SD = 0.80) followed by emotional dysregulation (M = 3.33, SD = 0.75), social stress (M = 3.29, SD = 0.74), suicide risk indicators (M = 3.16, SD = 0.83), and impulsivity (M = 3.07, SD = 0.78). Reliability analysis confirmed strong internal consistency across all measurement scales, with Cronbach’s alpha values ranging from 0.81 to 0.90. Multiple regression analysis indicated that the predictive model explained 58% of the variance in suicide risk outcomes (R² = 0.58). Depressive symptoms emerged as the strongest predictor of suicide vulnerability with a standardized beta coefficient of 0.42 (p < 0.001), followed by emotional dysregulation (β = 0.30, p < 0.001) and social stress (β = 0.27, p < 0.001), while impulsivity (β = 0.18, p = 0.009) demonstrated a moderate but statistically significant influence. These findings indicate that suicide risk among adolescents is shaped by a multidimensional interaction of emotional distress, behavioral tendencies, and environmental stressors. The integration of these psychodiagnostics indicators into AI-based predictive models may enhance early identification of vulnerable adolescents and support more effective suicide prevention strategies within mental health and educational systems.