healthliberal
Exploring Student Minds: A New Way to Spot Depression, Anxiety and Stress
BangladeshFriday, May 22, 2026
Researchers surveyed 424 students in July 2024, amid social and political tension. The findings reveal a striking prevalence of mental health symptoms:
- Depression: ~66 % reported symptoms
- Anxiety: >70 % felt anxious
- Stress: >50 % experienced stress
- Co‑occurrence: Nearly half reported all three simultaneously
Methodology
| Step | Technique | Purpose |
|---|---|---|
| Data split | Repeated training/testing partitions | Robust performance evaluation |
| Models tested | 8 machine‑learning algorithms | Identify best predictors |
| Best performers | Support‑Vector Machines (SVM) for depression & stress; XGBoost for anxiety | Highest predictive accuracy |
Explainability
- Tools: SHAP, LIME
- Key risk factors identified:
- Sleep disturbances
- Mental fatigue
- Personal behavior changes (general)
- Anxiety: Study habits, internet usage
- Depression: Gender, university type
Implications
- Not a diagnostic tool; does not replace clinical assessment.
- Demonstrates how explainable AI can uncover patterns in student mental‑health data, guiding future research priorities.
- Findings are specific to this sample and should not be generalized to all students.
Takeaway
Explainable machine‑learning offers a promising avenue for early identification of mental‑health risks among students, enabling targeted interventions while respecting the nuances of individual contexts.
Actions
flag content
