What does this research mean for the field?

A machine learning-powered mobile application can accurately detect stress in real-time using wearable electrodermal activity and skin temperature data and successfully deliver personalized mitigation interventions. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

What question did this study set out to answer?

The study aims to explore the integration of machine learning with wearable technology for stress detection and management.

June 7, 2026Open Access

An Intelligent Mobile Application for Real-Time Stress Detection and Customized Mitigation Techniques Using Wearable Physiological Data: A Preliminary Study

Key Points

The study aims to explore the integration of machine learning with wearable technology for stress detection and management.
Developed an intelligent mobile application for stress detection using wearable physiological data.
Utilized electrodermal activity and skin temperature signals for classification of stress states using machine learning models.
Collected feedback from five users on the system's usability and effectiveness of interventions.
The XGBoost model achieved an accuracy of 91.18% in stress detection.
The SVM model achieved an accuracy of 94.12% in classifying stress vs. rest states.
User feedback was positive regarding the system's comprehension and intervention effectiveness.

Abstract

Stress is a critical factor impacting human performance within transportation systems, often leading to cognitive fatigue, reduced situational awareness, and increased accident risks among operators and drivers. If unmanaged, chronic stress also results in serious health conditions such as hypertension and cardiovascular diseases. In the context of smart mobility, the early detection of stress is vital for maintaining both operator’s well-being and overall system safety. While wearable devices allow for the continuous collection of physiological signals, most existing detection and mitigation systems are overly complex and lack tailored intervention strategies for mobile environments. Hence, we propose an intelligent mobile app powered by machine learning models for real-time detection and management of stress using data obtained from wearable sensors. The proposed system classifies a subject as either stress or rest based on electrodermal activity (EDA) and skin temperature (TEMP) signals, using the XGBoost model trained on the open WESAD dataset. As soon as the subject is detected in a stress condition, the mobile application will trigger a personalized relaxation response comprising soothing music, motivational messages, or guided breathing exercises. In the experiments, the XGBoost model achieved an accuracy of 91.18 %, whereas the SVM model achieved an accuracy of 94.12%, demonstrating the strong capability of both models in distinguishing between stress and rest states. This study aimed to provide an initial exploration of the potential for integrating machine learning–based stress detection with wearable technologies, supported by personalized intervention methods. Feedback was collected from five users regarding the system’s ease of understanding and the effectiveness of the personalized interventions. The responses were positive and encouraging, supporting further development and enhancement of the system.

Read Full Paperexternally

Mark Helpful

Bookmark

Relay

View Full Paper