A Machine Learning Framework for Early Depression Screening Based On Daily Activities

DOI

10.2991/978-94-6239-664-7_27How to use a DOI?

Keywords

Depression Screening; Machine Learning Algorithms; University Students; SHAP Explanations; XGBoost Classifier; Mental Health Monitoring

Abstract

There is an increasing number of depressed university students in low- and middle-income communities, yet there are few large-scale, validated tools for screening for depression. Using a structured questionnaire, we followed an inquiry protocol that included questions on everyday habits, bedtime, screen time, socialization, physical activity, diet, and study time. These questions were administered to 1000 university students in Bangladesh. We categorized them along their prevalence in depression measured by the PHQ-9 test, by assigning them four levels of depression labels. After the preprocessing stage and applying a 2:1 train-test split, we applied logistic regression, SVM (RBF), multi-layer perceptron, and XGBOOST. We have used a grid search implementation of hyperparameter optimization and 10-fold cross-validation. We interpreted the results obtained by Lime SHAP and tested its performance using precision, recall, F1 score, and the area under the curve. XG Boost achieved a higher accuracy (86.9%) and AUC score, with the highest accuracy and F1 score, compared to Logistic Regression and SVM (p < 0.05). In addition to regular eating habits and exercise, SHAP showed that the quality of sleep, number of social interactions, hours spent on a screen before bedtime, and university workload were the main determinants. Interpretable gradient boosting on self-reported behaviors offers an accurate, transparent, and cost-effective approach to early depression screening in universities, even with limited resources. Further research should display mean values differently, since different demographic populations could exist. Moreover, advanced multimodal privacy-preserving implementations should be considered.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Rahat Ahmed
AU  - Mrinal Kanti Baowaly
PY  - 2026
DA  - 2026/06/08
TI  - A Machine Learning Framework for Early Depression Screening Based On Daily Activities
BT  - Proceedings of the International Conference on Intelligent Data Analysis and Applications (IDAA 2025)
PB  - Atlantis Press
SP  - 385
EP  - 396
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-664-7_27
DO  - 10.2991/978-94-6239-664-7_27
ID  - Ahmed2026
ER  -