Ansiedad en adolescentes: factores asociados identificados mediante modelos de aprendizaje automático en El Salvador

Abstract

Introduction. Anxiety in adolescence is a growing public health problem that affects emotional well-being, academic performance, and social relationships. Objective. Analyze the factors associated with the presence of anxiety symptoms in adolescents in El Salvador. Methodology. A cross-sectional analytical study was conducted with 1,209 adolescents using data from the 2022 National Mental Health Survey. Descriptive analysis, comparison of proportions, binomial logistic regression, and Random Forest were applied to datasets consisting of 80 % training and 20% test data. Model performance was evaluated using the area under the ROC curve and confusion matrices. Results. The prevalence of anxiety symptoms was 30.9 %. Being a student (OR = 4.67), experiencing personal stressors (OR = 3.23), and being female (OR = 2.29) were associated with a higher likelihood of anxiety symptoms. In contrast, the non-case categories for depression(OR = 0.06), post-traumatic symptoms (OR = 0.26), emotional symptoms (OR = 0.32), and suicide risk (OR = 0.52) were associated with a lower probability of the outcome, p < 0.05. The models demonstrated favorable discriminatory power (average AUC = 0.86 for RL and 0.87 for RF). Conclusion. Anxiety symptoms in adolescents result from the interaction of individual, school-related, and family factors. The integration of logistic regression and machine learning provides useful evidence to guide future research and comprehensive mental health strategies for adolescents.