Iranian Association for Intelligence and Talent Studies International Journal of Education and Cognitive Sciences 3041-8828 7 September 2026 2026 09 01 Application of Artificial Intelligence Models to Explore the Optimal Structural/Predictive Model of Gender Transition Request Based on Childhood Trauma Application of Artificial Intelligence Models to Explore the Optimal Structural/Predictive Model of Gender Transition Request Based on Childhood Trauma 1 14 10.61838/kman.ijecs.344 EN Journal Article 2025 10 01 <p><strong>Purpose: </strong>This study aimed to develop and evaluate an integrative structural–predictive model combining structural equation modeling and artificial intelligence to identify key trauma-related and psychosocial predictors of requests for gender transition among Iranian adolescent girls.</p> <p><strong>Methods and Materials: </strong>The study employed a descriptive–analytical, causal–comparative design with a sample of 300 adolescent girls aged 13–18 years (150 requesting gender transition and 150 controls) recruited in Tehran. Standardized measures of childhood trauma, general mental health, psychosocial functioning, and perceived stress were administered. Structural equation modeling was conducted using AMOS to test hypothesized causal pathways, followed by training and evaluation of decision tree, random forest, support vector machine, and fuzzy neural network models using Python.</p> <p><strong>Findings: </strong>Structural equation modeling demonstrated significant direct and indirect effects of childhood trauma on psychological disorders, mediated by psychosocial factors and perceived stress (p &lt; 0.001), with high explained variance for mental health outcomes. Among artificial intelligence models, the hybrid fuzzy neural network achieved the highest performance (accuracy = 95.6%, sensitivity = 97.8%, AUC–ROC = 0.98), significantly outperforming logistic regression and standalone models. Interpretability analyses identified sexual abuse and depression as the strongest predictors contributing to model decisions.</p> <p><strong>Conclusion: </strong>The findings support a trauma-informed, stress-based conceptualization of gender dysphoria in adolescent girls and demonstrate that integrating theory-driven structural modeling with artificial intelligence yields both high predictive accuracy and clinical interpretability, offering a robust framework for early identification and intervention.</p> https://journalecs.com/index.php/ecs/article/download/344/307