<p>Juvenile depression poses a critical public health challenge due to its high prevalence and strong association with youth suicide. However, current detection methods lack the sensitivity and specificity required for reliable early identification, hindering timely intervention and increasing the risk of adverse outcomes. This study addresses this gap by developing an artificial intelligence model based on machine learning, designed to enhance the accuracy of depression diagnosis in young people using sociodemographic and clinical variables. Using the PHQ-9 questionnaire, specific questions —such as the frequency of self-harm thoughts, lack of interest in activities, concentration problems, and low self-esteem— were identified as highly contributive to depression classification. Additionally, educational level and area of residence showed a significant correlation: young people with secondary education and those in urban areas had higher probabilities of severe depressive symptoms. Models such as Support Vector Machines (SVM), K-Nearest Neighbors (KNN), and Random Forest (RF) were compared, with RF proving to be the most effective, achieving 87% accuracy with low metric variability. Unlike deep neural networks, which require high computational resources, machine learning models like RF provide a more cost-effective solution by employing less complex algorithms and demanding fewer processing and memory resources, enabling implementation in resource-limited settings. This approach highlights the potential of machine learning models to improve early detection of depression in young populations, facilitating more effective and accessible intervention for vulnerable groups.</p>