A Behavioral Data-Driven Machine Learning Framework for Real-Time Academic Stress Prediction in University Students
- Version
- Download 31
- File Size 456.39 KB
- File Count 1
- Create Date 5 March 2026
- Last Updated 5 March 2026
A Behavioral Data-Driven Machine Learning Framework for Real-Time Academic Stress Prediction in University Students
Rasika R. Jadhao1, Dr. Brijendra Gupta2
1,2 Department of Information Technology, Siddhant College of Engineering, Savitribai Phule Pune University, Pune, Maharashtra, India
Abstract - Academic stress is a pervasive challenge in higher education, significantly affecting student’s mental health, learning efficiency and long-term outcomes. Existing stress assessment approaches rely primarily on self-reported surveys or retrospective analysis, which are subjective and unable to capture rapid temporal fluctuations. This work presents a behavioural data-driven machine learning framework for near real-time academic stress prediction using passive smartphone sensing.The proposed framework continuously captures digital behavioural biomarkers such as screen interaction patterns, mobility dynamics, communication activity and daily routine regularity. These signals are transformed into interpretable behavioural features and analysed using robust ensemble learning models to infer stress states with high temporal resolution. Unlike conventional approaches, the framework enables early detection of stress escalation without requiring wearable sensors or frequent user input.Experimental evaluation on longitudinal behavioural data collected over an academic semester demonstrates that ensemble models significantly outperform traditional classifiers, achieving high multi-class prediction accuracy. Importantly, the system identifies stress escalation trends 24–72 hours before reported stress peaks. The findings confirm that passive behavioural signals function as reliable indicators of academic stress dynamics. This study contributes a scalable, non-intrusive and ethically grounded solution for proactive mental health monitoring and advances the field of digital mental health analytics.
Key Words: Machine Learning, Academic Stress, Passive Sensing, Smartphone Sensing, Real-Time Monitoring, Ensemble Learning, Predictive Modelling.
Download