Zizheng Guo
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zizheng Guo has a total of 45 co-authors. Here are the top co-authors by number of shared publications:
- Zhenqi Chen
School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China
4 shared publications - Guofa Li
Chongqing University
3 shared publications - Jingyu Zhang
State Key Laboratory of Cognitive Science and Mental Health, Institute of Psychology, Chinese Academy of Sciences, Beijing 100000, China; Department of Psychology, University of Chinese Academy of Sciences, Beijing 100000, China
2 shared publications Jun Zhang
School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China
2 shared publications- Qiaofeng Guo
School of Transportation and Logistics, Southwest Jiaotong University, Chengdu 610031, China
2 shared publications - Yufan Pan
School of Information Science and Technology, Southwest Jiaotong University, Chengdu, China
2 shared publications
Latest Publications
- Open Access
- 13 October 2026
- Published by Elsevier in SSRN Electronic Journal
Hyperscanning measurement of directional inter-brain information flow and communication-neural alignment in railway dispatcher-station attendant collaboration
Safety-critical collaboration requires measures that characterize how information is organized between operators during task execution rather than relying solely on task outcomes. This study developed a directional EEG hyperscanning framework to measure dyadic collaboration between railway dispatchers and station attendants. Three communication conditions were examined: text from dispatcher to station attendant (Text D→S), voice from dispatcher to station attendant (Voice D→S), and voice from station attendant to dispatcher (Voice S→D). Partial directed coherence (PDC) was used to estimate directional inter-brain connectivity. A directional flow index (DFI) quantified the overall asymmetry of inter-brain information flow, while communication-neural alignment (CNA) related neural flow direction to the communication direction required by the task. Text D→S showed stronger D→S directional asymmetry than both voice conditions across the theta, alpha, and beta bands. In contrast, DFI did not differ significantly between the two voice conditions, indicating that reversing voice communication was not accompanied by a corresponding reversal in inter-brain information flow. At the trial level, the association between DFI and collaborative success varied with communication condition, whereas higher within-condition alpha-band CNA was associated with a greater probability of collaborative success. These findings show that directional inter-brain information flow cannot be interpreted solely from its absolute direction but needs to be considered in relation to task-level communication direction. The proposed framework provides a task-referenced, process-level approach for measuring the directional organization of safety-critical collaboration beyond conventional outcome-based assessment.
- 13 October 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessHyperscanning measurement of directional inter-brain information flow and communication-neural alignment in railway dispatcher-station attendant collaboration
Safety-critical collaboration requires measures that characterize how information is organized between operators during task execution rather than relying solely on task outcomes. This study developed a directional EEG hyperscanning framework to measure dyadic collaboration between railway dispatchers and station attendants. Three communication conditions were examined: text from dispatcher to station attendant (Text D→S), voice from dispatcher to station attendant (Voice D→S), and voice from station attendant to dispatcher (Voice S→D). Partial directed coherence (PDC) was used to estimate directional inter-brain connectivity. A directional flow index (DFI) quantified the overall asymmetry of inter-brain information flow, while communication-neural alignment (CNA) related neural flow direction to the communication direction required by the task. Text D→S showed stronger D→S directional asymmetry than both voice conditions across the theta, alpha, and beta bands. In contrast, DFI did not differ significantly between the two voice conditions, indicating that reversing voice communication was not accompanied by a corresponding reversal in inter-brain information flow. At the trial level, the association between DFI and collaborative success varied with communication condition, whereas higher within-condition alpha-band CNA was associated with a greater probability of collaborative success. These findings show that directional inter-brain information flow cannot be interpreted solely from its absolute direction but needs to be considered in relation to task-level communication direction. The proposed framework provides a task-referenced, process-level approach for measuring the directional organization of safety-critical collaboration beyond conventional outcome-based assessment.
- Open Access
- 7 September 2026
- Published by Elsevier in SSRN Electronic Journal
Ordinal Transition-Aware EEG-Based Detection of Mind Wandering in High-Speed Rail Drivers
Mind wandering during high-speed rail (HSR) operation may impair drivers’ sustained attention and increase safety risks. To effectively detect mind wandering in professional HSR drivers, this study proposes a region-aware ordinal transition-aware Transformer-LSTM framework based on EEG signals. Twenty professional HSR drivers participated in a simulated driving experiment, and the thought-probe method was used to label cognitive states as on-task, mild mind wandering, or deep mind wandering according to a 9-point Mind Wandering Scale. Power spectral density, band-power ratio, and Shannon entropy features were extracted from 62-channel EEG signals. A two-stage feature selection method combining mutual information and permutation testing was then used to reduce feature redundancy and identify discriminative EEG patterns with regional and spectral interpretability. The selected features were modeled using a region-aware Transformer encoder to capture global spatial-spectral dependencies, followed by an LSTM module to characterize temporal state dynamics. Across five repeated stratified trial-level evaluations, the proposed model achieved an accuracy of 87.20% ± 0.50%, a macro-F1 score of 87.12% ± 0.51%, and a weighted-F1 score of 87.21% ± 0.50%, outperforming several widely used models and enabling effective detection of mind wandering in HSR drivers. Ablation results confirmed the contributions of region-aware representation, ordinal and transition-aware constraints, and Transformer-LSTM temporal modeling. Further limited-sensor analysis showed that parieto-occipital EEG features retained substantial discriminative information, suggesting the feasibility of simplified wearable EEG deployment. These findings demonstrate the potential of EEG-based three-level mind-wandering detection for intelligent railway safety monitoring and early-warning systems.
- 7 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessOrdinal Transition-Aware EEG-Based Detection of Mind Wandering in High-Speed Rail Drivers
Mind wandering during high-speed rail (HSR) operation may impair drivers’ sustained attention and increase safety risks. To effectively detect mind wandering in professional HSR drivers, this study proposes a region-aware ordinal transition-aware Transformer-LSTM framework based on EEG signals. Twenty professional HSR drivers participated in a simulated driving experiment, and the thought-probe method was used to label cognitive states as on-task, mild mind wandering, or deep mind wandering according to a 9-point Mind Wandering Scale. Power spectral density, band-power ratio, and Shannon entropy features were extracted from 62-channel EEG signals. A two-stage feature selection method combining mutual information and permutation testing was then used to reduce feature redundancy and identify discriminative EEG patterns with regional and spectral interpretability. The selected features were modeled using a region-aware Transformer encoder to capture global spatial-spectral dependencies, followed by an LSTM module to characterize temporal state dynamics. Across five repeated stratified trial-level evaluations, the proposed model achieved an accuracy of 87.20% ± 0.50%, a macro-F1 score of 87.12% ± 0.51%, and a weighted-F1 score of 87.21% ± 0.50%, outperforming several widely used models and enabling effective detection of mind wandering in HSR drivers. Ablation results confirmed the contributions of region-aware representation, ordinal and transition-aware constraints, and Transformer-LSTM temporal modeling. Further limited-sensor analysis showed that parieto-occipital EEG features retained substantial discriminative information, suggesting the feasibility of simplified wearable EEG deployment. These findings demonstrate the potential of EEG-based three-level mind-wandering detection for intelligent railway safety monitoring and early-warning systems.
- 14 February 2026
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 230, 108455
Railway drivers’ physiological responses to typical hazardous scenarios: differences between professional drivers and student drivers
- 14 February 2026
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 230, 108455
Railway drivers’ physiological responses to typical hazardous scenarios: differences between professional drivers and student drivers
- 21 November 2023
- Published by Springer Nature in Cognition, Technology & Work
- Vol. 26 (1) , 153-168
The influence of train driver's foreign body penetration experience on hazard perception sensitivity: the mediating role of sustained attention
- 21 November 2023
- Published by Springer Nature in Cognition, Technology & Work
- Vol. 26 (1) , 153-168
The influence of train driver's foreign body penetration experience on hazard perception sensitivity: the mediating role of sustained attention
- 4 October 2023
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 193, 107332
A multi-level analysis on the causes of train-pedestrian collisions in Southwest China 2011–2020
- 4 October 2023
- Published by Elsevier in Accident Analysis & Prevention
- Vol. 193, 107332
A multi-level analysis on the causes of train-pedestrian collisions in Southwest China 2011–2020
- 10 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Radio Frequency Identification
- Vol. 6, 825-831
Driver Distraction Detection Using Advanced Deep Learning Technologies Based on Images
Driver distraction is an important factor leading to traffic accidents. The identification of drivers’ distracted states is of great significance to improve traffic safety. This paper presents a new driver distraction discrimination method by comprehensively deploying transfer learning technologies in deep learning algorithms. We use three publicly available datasets (i.e., state farm, origin, and 3MDAD) to verify the effectiveness of the developed methods based on Alexnet, VGG16, and Resnet18. The images with high recognition errors are further investigated by class activation map. The experiment results show that Resnet18 on origin has the best performance. The accuracies of the developed methods based on Alexnet, VGG16, and Resnet18 can achieve 99.92%, 100%, and 99.99% on the origin dataset, respectively. The accuracies of the Resnet18 -based method on state farm and 3MDAD are 84.82% and 97.7%, respectively. Our proposed methods can well support the development of distraction detection applications in transportation systems.
- 10 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Radio Frequency Identification
- Vol. 6, 825-831
Driver Distraction Detection Using Advanced Deep Learning Technologies Based on Images
Driver distraction is an important factor leading to traffic accidents. The identification of drivers’ distracted states is of great significance to improve traffic safety. This paper presents a new driver distraction discrimination method by comprehensively deploying transfer learning technologies in deep learning algorithms. We use three publicly available datasets (i.e., state farm, origin, and 3MDAD) to verify the effectiveness of the developed methods based on Alexnet, VGG16, and Resnet18. The images with high recognition errors are further investigated by class activation map. The experiment results show that Resnet18 on origin has the best performance. The accuracies of the developed methods based on Alexnet, VGG16, and Resnet18 can achieve 99.92%, 100%, and 99.99% on the origin dataset, respectively. The accuracies of the Resnet18 -based method on state farm and 3MDAD are 84.82% and 97.7%, respectively. Our proposed methods can well support the development of distraction detection applications in transportation systems.
- Open Access
- 3 September 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (17) , 11056
How Work-Family Conflict Influenced the Safety Performance of Subway Employees during the Initial COVID-19 Pandemic: Testing a Chained Mediation Model
This study examined the impact of work-family conflict on subway employees’ safety performance during the initial wave of the COVID-19 pandemic. We proposed a chain mediation model in which job burnout and affective commitment play mediating roles in this process. Using questionnaire data from 632 Chinese subway employees during February 2020, structural equation modeling analyses were performed. The analyses showed that work-family conflict had a significant negative impact on subway employee safety performance. Moreover, job burnout completely mediated the influence of work-family conflict on safety performance, while affective commitment only partially mediated the influence of job burnout on safety performance. These findings suggest the important role played by Work-Family balance during the pandemic and contribute to a deeper understanding of the inner mechanisms. We also discussed several practical implications for organizations to reduce the negative impact of work-family conflict on safety performance.
- 3 September 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (17) , 11056
Open AccessHow Work-Family Conflict Influenced the Safety Performance of Subway Employees during the Initial COVID-19 Pandemic: Testing a Chained Mediation Model
This study examined the impact of work-family conflict on subway employees’ safety performance during the initial wave of the COVID-19 pandemic. We proposed a chain mediation model in which job burnout and affective commitment play mediating roles in this process. Using questionnaire data from 632 Chinese subway employees during February 2020, structural equation modeling analyses were performed. The analyses showed that work-family conflict had a significant negative impact on subway employee safety performance. Moreover, job burnout completely mediated the influence of work-family conflict on safety performance, while affective commitment only partially mediated the influence of job burnout on safety performance. These findings suggest the important role played by Work-Family balance during the pandemic and contribute to a deeper understanding of the inner mechanisms. We also discussed several practical implications for organizations to reduce the negative impact of work-family conflict on safety performance.
- Open Access
The Effect of Time Window Length on EEG-Based Emotion Recognition
Various lengths of time window have been used in feature extraction for electroencephalogram (EEG) signal processing in previous studies. However, the effect of time window length on feature extraction for the downstream tasks such as emotion recognition has not been well examined. To this end, we investigate the effect of different time window (TW) lengths on human emotion recognition to find the optimal TW length for extracting electroencephalogram (EEG) emotion signals. Both power spectral density (PSD) features and differential entropy (DE) features are used to evaluate the effectiveness of different TW lengths based on the SJTU emotion EEG dataset (SEED). Different lengths of TW are then processed with an EEG feature-processing approach, namely experiment-level batch normalization (ELBN). The processed features are used to perform emotion recognition tasks in the six classifiers, the results of which are then compared with the results without ELBN. The recognition accuracies indicate that a 2-s TW length has the best performance on emotion recognition and is the most suitable to be used in EEG feature extraction for emotion recognition. The deployment of ELBN in the 2-s TW can further improve the emotion recognition performances by 21.63% and 5.04% when using an SVM based on PSD and DE features, respectively. These results provide a solid reference for the selection of TW length in analyzing EEG signals for applications in intelligent systems.
Open AccessThe Effect of Time Window Length on EEG-Based Emotion Recognition
Various lengths of time window have been used in feature extraction for electroencephalogram (EEG) signal processing in previous studies. However, the effect of time window length on feature extraction for the downstream tasks such as emotion recognition has not been well examined. To this end, we investigate the effect of different time window (TW) lengths on human emotion recognition to find the optimal TW length for extracting electroencephalogram (EEG) emotion signals. Both power spectral density (PSD) features and differential entropy (DE) features are used to evaluate the effectiveness of different TW lengths based on the SJTU emotion EEG dataset (SEED). Different lengths of TW are then processed with an EEG feature-processing approach, namely experiment-level batch normalization (ELBN). The processed features are used to perform emotion recognition tasks in the six classifiers, the results of which are then compared with the results without ELBN. The recognition accuracies indicate that a 2-s TW length has the best performance on emotion recognition and is the most suitable to be used in EEG feature extraction for emotion recognition. The deployment of ELBN in the 2-s TW can further improve the emotion recognition performances by 21.63% and 5.04% when using an SVM based on PSD and DE features, respectively. These results provide a solid reference for the selection of TW length in analyzing EEG signals for applications in intelligent systems.
- Open Access
- 17 May 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (10) , 6104
Characteristics of Train–Pedestrian Collisions in Southwest China, 2011–2020
Although train–pedestrian collisions are the primary source of railway casualties, the characteristics of this phenomenon have not been fully investigated in China. This study examined such collisions in the Greater Sichuan-Chongqing area of China by conducting a thorough descriptive analysis of 2090 incident records from 2011 to 2020. The results showed that such collisions have declined gradually over the past decade, but the fatality rate remains high. We found that such collisions were more likely to happen to men, senior citizens and people crossing the tracks and that they occurred more frequently in the morning. While collision rates dropped in February, collisions were more likely to occur in December. In contrast to the situation in Western countries, weekends were not related to increased occurrence. The absence of a protective fence led to a higher collision rate, but level crossings are no longer a concern since most such structures in China have been rebuilt as overpasses. Mild slopes and extreme curvatures were also found to increase the occurrence of such collisions. Freight trains were most likely to be involved in train–pedestrian collisions, and collisions caused by high-speed trains were rare both absolutely and relatively. However, when collisions did occur, higher train speeds were linked with higher fatality rates. The findings suggest that patterns of train–pedestrian collisions in China differ from those in the Western world. This difference might be caused by differences in culture, geography, weather and railway development policies. Future research directions and possible preventive measures are also discussed.
- 17 May 2022
- Published by MDPI AG in International Journal of Environmental Research and Public Health
- Vol. 19 (10) , 6104
Open AccessCharacteristics of Train–Pedestrian Collisions in Southwest China, 2011–2020
Although train–pedestrian collisions are the primary source of railway casualties, the characteristics of this phenomenon have not been fully investigated in China. This study examined such collisions in the Greater Sichuan-Chongqing area of China by conducting a thorough descriptive analysis of 2090 incident records from 2011 to 2020. The results showed that such collisions have declined gradually over the past decade, but the fatality rate remains high. We found that such collisions were more likely to happen to men, senior citizens and people crossing the tracks and that they occurred more frequently in the morning. While collision rates dropped in February, collisions were more likely to occur in December. In contrast to the situation in Western countries, weekends were not related to increased occurrence. The absence of a protective fence led to a higher collision rate, but level crossings are no longer a concern since most such structures in China have been rebuilt as overpasses. Mild slopes and extreme curvatures were also found to increase the occurrence of such collisions. Freight trains were most likely to be involved in train–pedestrian collisions, and collisions caused by high-speed trains were rare both absolutely and relatively. However, when collisions did occur, higher train speeds were linked with higher fatality rates. The findings suggest that patterns of train–pedestrian collisions in China differ from those in the Western world. This difference might be caused by differences in culture, geography, weather and railway development policies. Future research directions and possible preventive measures are also discussed.