A Learning Agent for Stress Multi-Level Diagnostics, Personalised Stress Profiles and Interventions in the Work Context

Abstract

Background: Work-related stress affects 39% of Austrians, contributing to mental health issues like depression and burnout, driven by factors such as workload and lack of control. Objectives: The Relax project aims to develop a holistic stress management framework using continuous stress assessment and personalized interventions based on physiological, behavioral, emotional, and cognitive indicators. Methods: The study combines wearable sensors (e.g., Polar Verity Sense), psychological methods, and technical strategies, with a longitudinal design to assess the app’s usability and effectiveness. Results: Usability was hindered by technical issues, but stress data visualization was well-received. The Micro-Model aligned well with stress estimates, while the Macro-Model faced data limitations. Methodological constraints suggested further refinement. Conclusion: The study highlights the need for individualized, multimodal stress management, with future research focusing on system improvements, technical refinement, and validation through an anonymized dataset.