Predicting University Students’ Lifestyle Quality (CEVJU) Using Socio-Demographic and Contextual Factors: A Machine Learning Approach for Personalized HCI Interventions
Journal
Communications in Computer and Information Science
HCI International 2026 Posters
Date Issued
2026
Author(s)
Iñiguez Jiménez, Samuel Olegario
Vinueza Fernández, Israel Santiago
Type
text::conference output::conference proceedings::conference paper
Abstract
The transition to university life is a critical period for health-behavior adoption. This study evaluates the predictive capacity of five low-effort socio-demographic and contextual variables age, sex, ethnicity, parish type, and the Multidimensional Poverty Index (MPI) for classifying lifestyle risk using the University Student Lifestyle Questionnaire (CEVJU) in 370 Ecuadorian undergraduates. A Random Forest model (10-fold × 3-repeat CV) achieved AUC-ROC = 0.560 and accuracy of 71.8% against a No-Information Rate of 68.9% (+2.9 pp). Parish type was the strongest predictor, followed by age and MPI. The central finding is diagnostic: static socio-contextual variables alone have limited predictive power, motivating a tiered personalization framework for proactive HCI systems that progressively incorporates real-time behavioral signals. This framework constitutes the primary HCI contribution. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.