As global stress levels reach high record, a breakthrough in Explainable AI (XAI) is offering a more transparent way to predict psychological crises. A research team led by Femi T. Johnson a member of the Irish Computer Society (ICS) from the Federal University of Agriculture, Abeokuta (FUNAAB), Nigeria has developed DeepStressScan, an AI model for detecting and predicting human stress.
DeepStressScan marks a pivotal shift from “Black Box” algorithms to interpretable clinical tools utilizing a transparent, dual-layered approach for solving the “Outlier” problem common in many machine learning models. Traditional machine learning often fails in mental health because extreme, high-risk cases are statistically rare, causing models to overlook them as “noise.”
The research, published in the Cureus Journal of Computer Science and available at https://doi.org/10.7759/s44389-025-09820-4 synthesized multi-sourced datasets from Indonesia, Bangladesh, and South Korea. By integrating the DASS-21 psychological scale with real-world lifestyle metrics, the team identified three primary “Stress Triggers”
i. Sleep Fragmentation: The most volatile predictor of acute stress spikes.
ii. Social Withdrawal: A lead indicator for severe depressive anomalies.
iii. Digital Saturation: High screen-time correlation with anxiety-related outliers.
“The goal isn’t just accuracy; it’s trust,” suggests the study’s methodology. Unlike deep learning models that hide their logic, DeepStressScan provides Feature Importance rankings. This will allow clinicians discover exactly which lifestyle factor (be it a lack of physical exercise or a drop in social engagement) is driving a “High Risk” alert. By proving that high-accuracy stress prediction can be achieved through interpretable ensemble methods, Johnson and his team are paving the way for AI integration in digital wellbeing apps and telehealth platforms. The model’s ability to function across diverse cultural datasets makes it a scalable solution for global health initiatives.
As DFI gains traction, it promises to transform precision mental health, offering a more holistic, scalable, and proactive approach to well-being. For organizations, this means healthier teams, reduced absenteeism, and a more supportive work environment. We’re proud to be pioneering this innovation and welcome collaboration with ICS members interested in the future of mental health.
About the Lead Researcher
Femi Temitope Johnson is an AI specialist in the Department of Computer Science, Federal University of Agriculture, Abeokuta, Nigeria whose work spans plant pathology and human behavior. His “Dual-Forest” approach is currently being studied for broader applications in diagnostic medicine and anomaly detection in critical infrastructure.