Predicting College Student Mental Health Levels: A Machine Learning Approach Using Sociodemographic and Quality of Life Data

• Lucas J. L. Braz UFC
• Lucas S. Fonseca UFC
• C. Alexandre R. Fernandes UFC

Resumo

Worldwide, mental health (MH) issues among students, including anxiety and depression, are rising. Predictive models are crucial for early intervention. This study proposes a novel MH prediction model for college students using data mining (DM) and machine learning (ML) techniques. The model integrates sociodemographic data and Quality of Life (QoL) assessments, employing various classification and regression methods, alongside feature selection. Using a database of 880 students and data from the Mental Health Inventory (MHI), WHOQOL-Bref, and a sociodemographic questionnaire, the model achieved a high accuracy of 82.62%, with an R² of 0.7139. Furthermore, the study identifies key factors influencing MH prediction. This data-driven approach provides a valuable tool for identifying students needing support, potentially improving intervention strategies and campus well-being programs.