P-434 Next-Gen Stress Management: AI’s Revolutionary Role in Workplace Occupational Health

Abstract Introduction Occupational stress accounts for nearly half of all lost workdays, underscoring the urgent need for real-time, practical solutions. To address this, we introduced a non-intrusive, multimodal approach for detecting work-related stress, leveraging videoplethysmography and self-reported metrics to capture a comprehensive picture of stress dynamics in real-life settings. Methods In our study involving 28 participants over three months, we collected physiological data during their regular 8-hour workdays, supplemented by self-reported measures collected through a specially designed app. This integrative method facilitated the labeling of physiological data with perceived stress levels, subsequently aiding in the development of an accurate stress detection model. Results The refined model exhibits an accuracy that exceeds 90% and an F1 score above 92%, highlighting its potential in fostering a healthier work environment. Discussion Moving forward, we aim to couple this detection model with a personalized digital coaching module, focusing on user-centric recommendations. This innovative strategy, grounded in user preferences, seeks not only to alert individuals to immediate stress peaks but also offers tailored suggestions for quick stress relief, thereby nurturing sustained engagement and adherence. Conclusion Our initiative showcases the feasibility of ethical, unobtrusive stress monitoring, promising a pathway to a healthier workforce by averting potential stress episodes and reducing chronic stress conditions. Future efforts will concentrate on assessing the positive impacts and efficacy of implementing such a recommender system.