Exploring the relationship between miners’ physiological signals and safety behavior in four emergency scenarios in coal mines: A virtual reality study

This study explores the relationship between miners' physiological signals and their safety behavior in emergency scenarios, with the aim of enhancing early warning systems for occupational safety. A virtual reality-based experiment was conducted in which miners were exposed to four simulated emergency scenarios. Physiological signals—including heart rate variability (Mean IBI, RMSSD, PNN50, LF/HF ratio), skin conductance (SC), and blood oxygen saturation (SPO₂)—were continuously recorded alongside behavioral performance measures such as reaction time and accuracy. Statistical analyses, including the Friedman test, repeated measures ANOVA, correlation analysis, and multivariate regression, revealed significant differences between emergency and resting conditions. Key findings indicated that faster reaction times were associated with lower PNN50 and SPO₂ but higher SC. Memory accuracy declined when PNN50 decreased and SC and skin temperature (SKT) increased. Discrimination performance improved with higher Mean IBI and SPO₂, but deteriorated with elevated RMSSD, PNN50, or reduced SKT. A classification model achieved 78.3 % accuracy in predicting miners’ safety behavior competence levels based on physiological inputs. These findings suggest a strong association between physiological responses and behavioral performance during emergencies, supporting the development of real-time physiological monitoring systems to enhance miners' safety. Further research should examine causal relationships and optimize predictive modeling approaches.