---
title: "Identifying Key Subway Operation Risk Factors Using the 24Model, Apriori, and Complex Network Anal"
id: "plos-one-12-comprehensive-analysis-method-of-key-risk-factors-in-the-subway-operation"
canonical_url: "https://medichelpline.com/clinical-feed/plos-one-12-comprehensive-analysis-method-of-key-risk-factors-in-the-subway-operation"
content_type: "clinical_feed_article"
specialty: "General"
source_name: "PLOS ONE (Medicine)"
source_url: "https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0358549"
published_at: "2026-09-18T14:00:00.000Z"
evidence_level: "Journal Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Identifying Key Subway Operation Risk Factors Using the 24Model, Apriori, and Complex Network Anal
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/plos-one-12-comprehensive-analysis-method-of-key-risk-factors-in-the-subway-operation
- **Specialty:** [General](https://medichelpline.com/clinical-feed/general.md)
- **Primary Source:** PLOS ONE (Medicine)
- **Source URL:** [Original Journal Publication](https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0358549)
- **Published At:** 2026-09-18T14:00:00.000Z
- **Evidence Rating:** Journal Feed
## Executive GIST (TL;DR)
- The study addresses increasing operational complexity and risk in China’s urban rail transit, noting 361 operating lines, 12,160.77 km total length, and 6,651 stations by the end of 2024. It frames subway safety as critical to service reliability and socioeconomic stability. - Authors applied an integrated, data-driven approach combining the **24Model** causation framework, association rule mining (Apriori algorithm), and **complex network** theory to extract and analyze risk factors from accident reports. - A corpus of 76 Chinese subway operation accident reports was coded using the 24Model to identify discrete risk factor nodes for subsequent analysis. - The Apriori algorithm produced association rules among extracted risk factors; these associations were used to construct a risk-factor network that represents couplings and interactions among causes. - Network robustness analysis and mutual information methods were applied to validate the network and to rank the importance of factors within the network structure. - Four highest-impact risk factors identified were: unclear safety responsibilities of employees; safety responsibility of managers; insufficient safety oversight of subcontractors; and lack of targeted content in safety training. - The paper argues that conventional safety management emphasizes isolated risk control and post-accident investigation, whereas the integrated network approach reveals system-level coupling and helps prioritize prevention and resource allocation. - The findings are presented as actionable evidence to support safety management decisions and the sustainable operation of subway systems. Data and supporting information are reported as available within the manuscript and its supporting files.
## Clinical Analysis & Structured Key Points
Comprehensive analysis method of key risk factors in the subway operation accident by complex network and accident data | PLOS One Browse Subject Areas ? Click through the PLOS taxonomy to find articles in your field. For more information about PLOS Subject Areas, click here . Article Authors Metrics Comments Media Coverage Reader Comments Figures Figures Abstract With the continuous expansion of subway systems in China, the operational systems of subway have become increasingly complex, leading to a rise in risk factors. In the event of an accident, these risks can significantly impact the safety and health of individuals, and pose a direct threat to the reliability, social stability, and economic sustainability of subway systems. To effectively prevent the occurrence of operational accidents and ensure the sustainable development of subway systems, it is essential to explore the key risk factors of subway accidents through appropriate technological approaches. This study proposes an integrated approach combining the 24Model, association rule mining, and complex network theory to conduct in-depth mining and analysis of textual data from accident reports, thereby identifying risk factors and exploring the coupling relationships and importance among them. First, risk factors were extracted by analyzing 76 reports using the 24Model. Then, the Apriori algorithm was applied to derive association rules among the risk factors, based on which a risk factor network model was constructed. Finally, the robustness analysis and mutual information theory were employed to validate the model and identify the key risk factors. The results show that unclear safety responsibilities of employees, safety responsibility of managers, insufficient safety oversight of subcontractors, and lack of targeted content in safety training are the four most critical risk factors. The findings of this study provide important safety management decision-making support for the development of more sustainable subway operational systems. Citation: Ma S, Jiang W (2026) Comprehensive analysis method of key risk factors in the subway operation accident by complex network and accident data. PLoS One 21(9): e0358549. https://doi.org/10.1371/journal.pone.0358549 Editor: Guanying Huang, City University of Hong Kong, HONG KONG Received: January 26, 2026; Accepted: September 2, 2026; Published: September 18, 2026 Copyright: © 2026 Ma, Jiang. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: All relevant data are within the manuscript and its Supporting Information files. Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. 1. Introduction In recent years, China’s urban rail transit has experienced rapid and large-scale development, with the length of operational lines ranking first globally for several consecutive years and reaching a world-leading level. By the end of 2024, a total of 361 urban rail transit lines were in operation across 58 cities in mainland China (hereafter referred to as China), covering a cumulative length of 12,160.77 kilometers and serving 6,651 stations, as illustrated in Fig 1 . China’s urban rail transit system has entered a period of rapid development; however, safety incidents and operational failures continue to occur frequently during metro operations [ 1 ]. Once an operational accident occurs, it may not only disrupt train services but also lead to large-scale passenger congestion, equipment damage, and substantial socioeconomic losses. Therefore, ensuring the safe operation of urban rail transit has become an essential component of the high-quality development of China’s urban rail transit sector. Download: PNG larger image TIFF original image Fig 1. Overview of urban rail transit development in China. The left panel presents the numbers of cities and operating lines and the annual growth rate of operating lines. The right panel presents the total length of operating lines, the number of operating stations, and their corresponding annual growth rates. https://doi.org/10.1371/journal.pone.0358549.g001 With the continuous expansion of subway network, the system has become increasingly complex, resulting in heightened uncertainty and vulnerability during operation. Given the numerous subsystems, high passenger mobility, and multifaceted risk factors, the subway system is influenced not only by internal dynamics but also by external disturbances. Consequently, any incident can significantly disrupt system stability and pose serious threats to the safety of urban residents and their property. For instance, on July 20, 2021, a catastrophic rainstorm caused severe flooding in Zhengzhou Metro Line 5, leading to the deaths of 14 passengers [ 2 ]; On June 22, 2009, nine people were killed and about 80 injured in a subway crash in Washington [ 3 ]. To improve the operational safety of urban rail transit, China has implemented a series of measures, including the continuous improvement of operational safety regulations and technical standards, the establishment of risk classification and control systems together with hazard identification and mitigation mechanisms, and the strengthening of safety training and job responsibility management for operational personnel. Meanwhile, some metro operators have begun to adopt intelligent monitoring technologies to dynamically monitor the operating status of trains, signaling systems, power supply systems, communication systems, and platform facilities. These measures have played a positive role in standardizing safety management practices and preventing equipment failures. However, existing safety management approaches still primarily focus on individual risk control, hazard identification, and post-accident cause investigation, while insufficient attention has been paid to the complex interrelationships and coupling effects among different risk factors. In particular, when numerous accident causative factors interact in a complex manner, conventional analytical methods have difficulty identifying the key risk factors with substantial influence and connectivity from a system-wide perspective. Furthermore, accident investigation reports in China are prepared under a relatively unified management framework, statistical methodology, and accident classification system, resulting in publicly available accident records that provide a relatively complete and consistent data foundation. In contrast, considerable differences exist among countries in terms of accident reporting systems, safety regulatory frameworks, accident classification standards, and data availability. Directly integrating accident data from multiple countries may therefore introduce substantial data heterogeneity, thereby affecting the reliability of risk-factor network construction and subsequent analyses. Therefore, based on subway operation accident data from China and complex network theory, this study constructs a risk-factor association network for subway operation accidents to identify key risk factors from the perspectives of network structure and factor interactions. The findings are expected to provide a more systematic basis for risk prevention and control as well as the allocation of safety management resources. Current research on subway operation accidents primarily focuses on causation analysis and risk assessment. In terms of causation analysis, The existing research can be categorized into two perspectives: the engineering and technical perspective and the data-driven perspective. The engineering and technical perspective focuses on direct physical causes of accidents. For example, Yao et al. [ 4 ] employed the finite element method to investigate the factors influencing passenger injuries in subway collision accidents. In contrast, the data-driven perspective aims to uncover systemic risk patterns from historical accident data. For example, Wang et al. [ 5 ], using accident case data, applied semantic networks and word frequency statistics to identify key risk factors in subway operation accidents. Zhang et al. [ 6 ] conducted statistical analyses of accident data to examine the patterns of subway incidents in Shanghai. In addition, network models have also been adopted to study subway accidents, primarily from the perspective of network topology structure [ 7 ]. However, these approaches face limitations in analyzing complex human and organizational factors, and they are inadequate for uncovering the deeper causes of subway operation accidents. Causation models, as an important tool for accident analysis, offer a systematic and multidimensional means of identifying critical factors involved in accidents, which can, to some extent, compensate for the aforementioned limitations. They provide a more comprehensive understanding of accident mechanisms and have been widely applied in the analysis of hazardous chemical accidents [ 8 – 10 ], coal mine disasters [ 11 – 13 ], power system failures [ 14 – 16 ], and construction site accidents [ 17 – 19 ]. In the field of subway operational accident research, He et al. analyzed the sources of operational risks using accident models such as fault tree analysis, and further identified rolling stock system failures and signal-communication system failures as the most critical risk factors through data envelopment analysis [ 20 ]. Wang et al. identified risk factors associated with subway operation accidents using accident models such as AcciMap, thereby providing support for constructing an accident semantic network [ 4 ]. As such, causation models are gradually emerging as a key method in the study of subway operation accidents. Therefore, this study adopts a causation model approach to identify risk factors in subway operation accidents. In the field of risk research, Wang et al. [ 7 ] developed an subway risk network model based on accident case studies. Derrible and Kennedy [ 21 ] investigated the vulnerability of network lines across 33 global metro systems using complex network theory. Lee and Hur [ 22 ] applied numerical simulation methods to analyze the risk of subway fire accidents. Complex network theory has been increasingly utilized in the study of risk evolution within subway networks [ 23 – 25 ], and its applications have extended to the analysis of risks in electrical accidents [ 26 , 27 ], construction incidents [ 28 , 29 ], and chemical industry accidents [ 30 , 31 ]. Compared to traditional risk assessment methods such as fault tree analysis and event tree analysis, complex networks offer distinct advantages in visualizing system structures, analyzing the intricate coupling relationships among risk factors, and identifying critical risk elements. Given that subway systems are inherently complex and their operational safety is influenced by a multitude of interconnected and interdependent factors, it is essential to explore the coupling mechanisms among various risk factors contributing to operational accidents. Such an investigation can help clarify the underlying formation mechanisms of these accidents. This, in turn, will support the development of more targeted and effective risk management strategies for subway operations. Therefore, this study employs complex network methods to identify key risk factors associated with subway operation accidents. Building upon the aforementioned research, this study integrates causation models, association rule mining, and complex network theory to construct a comprehensive analytical framework for identifying risk factors in subway operation accidents. The proposed model aims to uncover the key risk factors contributing to such accidents, thereby providing a theoretical foundation for enhancing the safety management of subway operations. A safe and reliable subway operational system is the prerequisite and cornerstone for achieving the sustainability goals of urban public transportation, such as environmental friendliness, economic efficiency, and social inclusiveness. This study, through data-driven analysis of key risk factors, aims to enhance the inherent resilience of the operational system. This is of significant importance for ensuring the long-term safety and stability of subway operations and promoting the sustainable development of subway systems. 2. Materials and methods 2.1. Data sources Accident cases can reveal the coupling mechanisms among risk factors and the evolutionary patterns of accidents, providing critical data support for risk factor analysis. To enhance the quality of accident data and ensure comprehensiveness and diversity, this study collected 76 subway operation accident cases that occurred in China between 2009 and 2024. The geographical and temporal scope of this study was determined by considering the comparability of accident data, the development trajectory of China’s urban rail transit, and data completeness. Restricting the analysis to China helps maintain the comparability of accident cases under relatively consistent accident-reporting practices, regulatory frameworks, and accident classification standards, thereby reducing data heterogeneity that could affect risk-factor identification and network construction. Previous studies have shown that China’s urban rail transit entered a period of rapid and large-scale development after 2008, with substantial growth in operating scale and the number of cities with operational systems during 2008–2015 [ 32 ]. After 2015, an increasing number of cities gradually transitioned from single-line operation to multi-line network operation, accompanied by continued advances in system automation and intelligence. Therefore, the period from 2009 to 2024 covers an important development process of China’s urban rail transit from rapid system expansion to large-scale networked operation, providing a broad empirical basis for identifying accident risk factors and their associations under different stages of operational development and operating environments. The year 2024 was selected as the endpoint because it was the most recent complete calendar year for which relatively comprehensive accident information was available when data collection was conducted; using a complete calendar year also helps reduce the risk of data truncation and sample omission caused by incomplete disclosure of more recent accident information. The cases were gathered from multiple sources, including official websites of provincial and municipal governments of the People's Republic of China, emergency management department portals, news reports, and published books. The accidents collected in this study occurred across 16 cities in China. Among them, the top three cities with the highest number of accidents—Beijing, Shenzhen, and Guangzhou—reported a total of 36 incidents, accounting for 47.37% of all cases. According to national standards and regulatory documents, including the Classification and Coding of Production Safety Accidents [ 33 ] and the Administrative Measures for Information Reporting and Analysis of Urban Rail Transit Operational Hazardous Incidents [ 34 ], the accident case database covered 13 categories of accident types, including train collision, train derailment, train conflict, vehicle-related injury, falls from height, and electric shock. In addition, the database also included severe accident consequence types such as vehicle damage, operation interruption, train delay, and casualties. The distribution of the specific accident types among the 76 accidents is presented in Table 1 . Download: PNG larger image TIFF original image Table 1. Distribution of the 76 subway operation accidents by accident type. https://doi.org/10.1371/journal.pone.0358549.t001 Although the 76 accident cases used in this study may not constitute a large sample size, they remain reasonably adequate within the framework of the present research. First, subway operation accidents are characterized as low-frequency but high-consequence events. Unlike high-incidence categories such as construction accidents and road traffic accidents, subway systems possess a relatively high degree of safety redundancy. Consequently, the number of accessible accident cases that can support in-depth coding and analysis based on the 24Model is inherently limited, and this study collected such cases to the greatest extent possible. Second, this study adhered to the principle of prioritizing data quality over quantity. The 24Model requires accident reports to contain sufficiently detailed information regarding the accident process, causation mechanisms, and related circumstances to support accident causation analysis. Blindly expanding the sample size by including incomplete cases, such as incidents documented only through brief news reports, could introduce substantial noise and thereby compromise the accuracy of data extraction. Despite the limited sample size, previous studies with comparable datasets have demonstrated the applicability and validity of similar approaches. For example, Su et al. [ 35 ] employed the Apriori algorithm based on 72 coal mine accidents to reveal the association relationships among coal mine safety risk factors and identify key risk factors. Wang et al. [ 36 ] extracted 418 strong association rules from 106 hazardous chemical accidents using the Apriori algorithm and subsequently utilized these rules as the topological structure of a Bayesian network model. Wu et al. [ 37 ] investigated the interaction relationships among subway construction safety risk factors based on 101 subway construction accident reports using association rule mining and complex network methods. Therefore, the sample size adopted in this study can be considered reasonably acceptable for supporting the methodological framework of the present research. 2. 2. Research framework The proposed methodology consists of three main components, as illustrated in Fig 2 . The first component involves the extraction of risk factors. Based on accident reports, risk factors are identified using the 24Model in combination with relevant laws, regulations, and technical standards. The second component focuses on the construction of a complex network. Using the established accident case database, association rule mining is applied to determine the relationships among risk factors. Risk factors are represented as network nodes, and their associations are represented as edges, thereby constructing the network's topological structure. The third component is complex network analysis. Key risk factors are identified through various analytical metrics, including node degree, node strength, and mutual information theory. Download: PNG larger image TIFF original image Fig 2. Research framework. The framework includes three stages: risk-factor extraction using the 24Model and relevant regulatory documents, complex-network construction through association rule mining, and identification of key risk factors using network topology indicators and mutual information. https://doi.org/10.1371/journal.pone.0358549.g002 2.3. Accident causation model Accident causation models are theoretical frameworks distilled from extensive accident investigations and analyses. These models reflect the underlying mechanisms of accidents and serve as essential tools for accident prevention, control, and root cause analysis. The 24Model is a theoretical framework for accident causation. It categorizes the causes of accidents into two hierarchical levels—organizational factors (including safety culture and safety management systems) and individual factors (including habitual behaviors, one-off behaviors, and physical conditions)—across four stages. This model is widely used f
## Related Clinical Research

- [Reinforced-count simulation: calibrating decisions under over-dispersed multi-type service demand](https://medichelpline.com/clinical-feed/plos-one-3-reinforced-count-simulation-for-decision-calibration-under-over-dispersed-multi.md)
- [Psychometric validation of the Patient-Centered Communication Scale (PCCS) in Iranian clinical nur](https://medichelpline.com/clinical-feed/plos-one-7-psychometric-features-of-the-patient-centered-communication-scale-among-iranian.md)
- [Drivers of patient satisfaction in Scottish general practice: deprivation, rurality and practice s](https://medichelpline.com/clinical-feed/bmj-open-13-patient-satisfaction-with-general-practice-in-scotland-secular-trends-and.md)
- [Accelerometer-derived sleep stages and incident disease risk in UK Biobank: phenome-wide cohort an](https://medichelpline.com/clinical-feed/plos-medicine-0-accelerometer-derived-real-world-sleep-stages-and-risk-of-incident-diseases-a.md)
- [Using Topologically Associated Domains to Prioritize Pathogenic Non-Coding Variants in Unresolved](https://medichelpline.com/clinical-feed/biorxiv-0-identifying-putative-pathogenic-non-coding-variants-in-unresolved-rare-disease.md)

## Navigation
- [← Back to General Feed](https://medichelpline.com/clinical-feed/general.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.