Experimental study on psychophysiological reactions to risk: identification of risk-taking behavior in virtual coal mining environment

View at publisher Summarise with VIRAT
Authors: Xiangchun Li, Yuzhen Long, Qin Li, Chunli Yang, Jianhua Zeng, S. Z. Zhang, Fangcai Jing
Cited by: 0
References: 94

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Purpose Unsafe behavior engaged by workers is a primary cause of coal mine accidents. This study aims to explore how to use psychophysiological signals to identify the risk-taking behavior of coal miners. Design/methodology/approach In this study, virtual reality technology was used to create experimental scenes of risk-taking behavior in coal mines, and ErgoLAB wireless wearable physiological recorders were used to measure the physiological signals of subjects. The changes of physiological signals were analyzed by multiple methods, such as box plot, normality test and significance tests. Three typical machine learning algorithms were used to develop the classification models of risk-taking behavior. A deep learning algorithm was used to analyze time-series data derived from critical indicators for the development of regression models characterizing emergency psychophysiological responses. Findings It was found that: skin conductance (SC), SC level and SC reaction significantly increased, while interbeat interval (IBI) and respiration range remarkably decreased during the risk-taking behavior; the classification models based on different modalities were developed by using K-nearest neighbor (KN), decision tree (DT) and random forest (RF) classifiers, with high accuracy scores exceeding 0.8; the Long Short-Term Memory (LSTM)-based regression models exhibited robust performance in characterizing physiological variations during emergency response. It was revealed that: IBI and SC serve as optimal indicators for the classification models; the RF classifier is more valid and robust for identifying risk-taking behavior than KN and DT; the LSTM algorithm is well-suited for capturing psychophysiological changes in emergency responses. Originality/value It is highly promising for the automatic identification of risk-taking propensities and the timely interventions of unsafe behaviors, ultimately preventing production safety accidents and reducing occupational injuries in the coal mining industry.

Fields of Study & Subjects

Industrial and Mining Safety Emotion and Mood Recognition Occupational Health and Safety Research

Publication details

Journal / source
Sensor Review
Pages
1–27
Publication date
28 March 2026
Language
EN