← Back to Publications List

Predicting Student Mental Health Scores Based on Social Media Usage Patterns Using Machine Learning

Students & Supervisors

Student Authors
Tohomina Rahman Tisha
Bachelor of Science in Computer Science & Engineering, FST
Hasin Almas Sifat
Bachelor of Science in Computer Science & Engineering, FST
Abida Afrin
Bachelor of Science in Computer Science & Engineering, FST
Ashiqur Rahman Saron
Bachelor of Science in Computer Science & Engineering, FST
Supervisors
Md. Mortuza Ahmmed
Associate Professor, Faculty, FST

Abstract

Research Area: This study focuses on the application of machine learning to predict mental health outcomes based on digital behavior analytics, particularly analyzing social media usage patterns among young adults. Objectives: The primary goal was to develop a robust predictive model capable of estimating students’ mental health scores from behavioral and demographic data collected via surveys. Methodology: The dataset included variables such as average daily social media use, sleep hours, academic performance, relationship status, and conflicts related to social media. After label encoding categorical features and standardizing the data, we trained an XGBoost Regressor model. Hyperparameters were optimized using grid search with 5-fold cross-validation. Results: The model achieved a high degree of accuracy, with a test R² score of 0.9425 and a cross-validated mean R² of 0.9674 (standard deviation 0.0110), demonstrating strong predictive performance and stability. Analysis: Feature importance analysis revealed that sleep hours, addiction score, average daily usage hours, conflicts over social media, and effects on academic performance were significant predictors of mental health scores. These results suggest the feasibility of using digital behavioral data for non-invasive mental health monitoring. Metric Value Mean Squared Error (MSE) 0.0691 R² Score (Test Set) 0.9425 Cross-Validated R² (Mean) 0.9674 Cross-Validated R² (Std Dev) 0.0110 Table 1: Performance of Model References [1] M. Drira, S. Ben Hassine, M. Zhang, and S. Smith, ""Machine Learning Methods in Student Mental Health Research: An Ethics-Centered Systematic Literature Review,"" Applied Sciences, vol. 14, no. 24, p. 11738, 2024. https://doi.org/10.3390/app142411738 Keywords: Mental Health, grid search, addiction score, XGBoost Regressor model "

Keywords

Keywords: Mental Health grid search addiction score XGBoost Regressor model

Publication Details

  • Type of Publication: Conference 
  • Conference Name: IEEE CS BDC Summer Symposium 2025
  • Date of Conference: 18/07/2025 - 18/07/2025
  • Venue: Hajee Mohammad Danesh Science and Technology University, Dinajpur-5200
  • Organizer: IEEE Computer Society Bangladesh Chapter
View on Publisher