Risk perception is known as a critical step in workers’ safety decision-making process. However, current approaches to assess workers’ perceived risk include surveys and interviews which are post hoc, subjective, and cumbersome to implement at construction sites. To address the issues associated with these methods, the authors propose a novel approach for the continuous and quantitative assessment of workers’ perceived risk using physiological responses acquired from wearable sensors. With this background, this study aims to investigate the potential of using physiological sensory data (e.g., electrodermal activity (EDA)) collected from off-the-shelf wristband typed sensors to understand construction workers’ perceived risk during their ongoing work. To achieve this objective, 30 h of physiological sensory data were collected from eight construction workers during their ongoing work. The results indicate that: (1) electrodermal response (EDR), which refers to short-term changes in EDA, shows significant differences between low and high-risk activities; (2) high-risk activities significantly affect workers’ EDR during their ongoing work. The main contribution of this study is to show the feasibility of using wearable sensors to understand workers’ perceived risk in construction sites continuously. Considering the complexity and dynamicity of workers’ tasks on construction sites, the development of an objective, continuous, and non-intrusive method for monitoring workers’ physiological responses is expected to contribute to a more in-depth understanding of construction workers’ perceived risk.
The authors would like to acknowledge their industry partners for their help in data collection, as well as anonymous participants who participated in the data collection. Also, the authors wish to thank Chris Soto from Skanska USA and Jad Ibrahim from Shade Cooperation for their support in collecting the field data and providing general feedback on workers\u2019 safety behaviors. The first author also wishes to acknowledge financial support by the University of Michigan from a Rackham Pre-doctoral Fellowship. All opinions and findings in this paper are those of the authors and do not necessarily represent those of the University of Michigan.