Abstract
This study presents the experimental findings investigating the effect of different interruptions on stress levels and validates indicators for interruption detection based on machine learning. Data were collected from one hundred and four subjects across simulated office-like tasks (i.e., writing and problem-solving) by four interruption types (i.e., distraction, intrusion, break, and multitasking). Behavioral data (RT and error rate), physiological signals (HR and EDR), and subjective ratings were recorded. Features based on ANOVA were extracted and entered into the classification models as inputs to examine how well different variables could be used to evaluate interruptions. Different stress outcomes were found on objective and subjective measures by interruption type. Also, we found that a combination of physiological signals, behavioral data, and subjective ratings provides the best accuracy. Additionally, the results suggest that task performance is a more reliable indicator than subjective ratings for interruption measurement and classification. This study uses methods that could be adapted for office workers, and the results provide preliminary support for developing an early interruption detection system. The methods employed in this research may benefit office staff, and the results offer preliminary support for developing an early interruption detection system to enhance worker health and company productivity.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Additional information
Notes on contributors
Elmira Zahmat Doost
Elmira Zahmat Doost is a postdoctoral associate in the Human System Engineering program at Arizona State University. Her research interests include human–computer interface, human factors psychology, cognitive ergonomics, human performance modelling, and behavioral technology simulation. She obtained her PhD in Management Science and Engineering from Tsinghua University, China.
Wei Zhang
Wei Zhang is a professor in Human Factors and Ergonomics in the Department of Industrial Engineering at Tsinghua University. He holds Master and PhD in Mechanical Engineering, Tsinghua University. His research areas are Driving Safety, Human–System Simulation, Virtual Reality and its applications.