Wearable devices for stress detection: Insights on exploring generalizability in controlled laboratory settings

Work-related stress in industrial environments presents significant challenges, affecting both employee well-being and organizational productivity. This study investigates the use of wearable devices for stress detection in controlled laboratory settings, focusing on the variability of stress induction protocols and physiological monitoring devices, such as the Empatica E4 and Zephyr BioHarness 3. Key physiological indicators, including heart rate variability and electrodermal activity, were analyzed, and a support vector machine algorithm was employed to classify stress levels with high accuracy. The primary contribution of this research lies in evaluating the adaptability of machine learning models across multiple stress protocols and device types, providing insights into their effectiveness in different controlled environments. These findings highlight the ongoing challenge of translating laboratory results to real-world industrial settings, where task complexity and environmental variability pose additional difficulties.