Analysis of depression in High School and Middle Basic adolescents using Machine Learning algorithms

García Saavedra Y.M.; Salinas Flores J.W.; García Saavedra M.D.P. · Psicogente (2024) · DOI: 10.17081/psico.27.52.6981

Resumen

Objective: To identify the determining factors associated with anxiety, stress, and depression in adolescents from sixth to eleventh grade at an educational institution in the department of Tolima, Colombia. Method: The Systemic Cognitive Inventory for the Study of Academic Stress (SISCO Inventory) was applied to a sample of 553 students aged 11 to 20 (mean = 17,12 years, SD = 0,766). Various machine learning classification algorithms were employed since the target variable was binary, complemented by ensemble methods. Results: When students were asked whether they had ever experienced persistent feelings of sadness, emptiness, or loss of interest in daily activities (indicative of depression), 76,17 % reported “yes,” while 23,83 % answered “no.” The probability of correctly detecting students with depression was 0,845, while for students without depression, it was 0,863. The main variables that best discriminated between students reporting depressive symptoms and those who did not include chronic fatigue, difficulty making decisions, anxiety, lack of appetite, interpersonal conflicts, evaluative activities, and drowsiness. Conclusions: The findings underscore the importance of early detection, the application of machine learning tools, and the need for preventive and personalized approaches to address anxiety, stress, and depression in adolescents.

Instituciones: Universidad Nacional Agraria La Molina