Identification of accident-injury type and bodypart factors from construction accident reports: A graph-based deep learning framework

Accident reports provide information to understand why and how events occur. Learning from past accident reports is critical for preventing accidents or injuries in construction safety management. However, there are two issues: (1) manual analysis of such accident reports is time-consuming and labor-intensive; and (2) previous research mainly focused on analyzing the causal factors of accidents. Not much research concentrates on the injury effect in an accident and the influential relationship between accident cause and injury effect. To tackle this problem, a graph-based deep learning framework is proposed to identify accident-injury type and bodypart factors automatically to enable managers to make timely and better-informed decisions to prevent accidents and injuries for on-site safety. In this framework, a graph-based deep learning approach (specifically, the Graph Convolutional Network) is developed to automatically classify accident reports labeled with accident_type and injury_type, whereas the traversal method is developed to identify the bodypart factors. To further intuitively visualize these safety risk factors (e.g., accident_type, injury_type, and bodypart factors), the co-occurrence networks are drawn to further intuitively reveal the interdependency in accident-injury and injury-bodypart types respectively. From the perspective of theoretical and practical contributions, the framework proposed in this study not only represents a substantial data-driven advancement in construction accident report classification and keyword extraction tasks, but also enables managers to get knowledge of construction safety performance (i.e., accident causes and injury effects) and further formulate corresponding strategies to prevent accidents and injuries in on-site safety management.     

Cognitive failure analysis for aircraft accident investigation

Abstract

The present studies were undertaken to investigate the applicability of an information processing approach to human failure in the aircraft cockpit. Using data obtained from official aircraft accident investigation reports, a database of accidents and incidents involving New Zealand civil aircraft between 1982 and 1991 was compiled. In the first study, reports were coded into one of three error stages proposed by Nagel (1988) and for the presence of any of 61 specific errors noted by Gerbert and Kemmler (1986). The importance of decisional factors in fatal crashes was noted. Principal components analysis suggested the presence of five different varieties of human failure. In the second study, a more detailed error taxonomy derived from the work of Rasmussen (1982) was applied to the data. Goal selection errors emerged as the most frequent kind of cognitive error in fatal accidents. Aircraft accident reports can be a useful source of information about cognitive failures if probed with an appropriate, theoretically-based, analysis of information processing errors. Such an approach could provide the accident investigators with a useful tool, and lead to a more complete understanding of human error in aviation.     

A methodology for identifying human error in U.S. Navy diving accidents

OBJECTIVE: To better understand how human error contributes to U.S. Navy diving accidents. BACKGROUND: An analysis of 263 U.S. Navy diving accident and mishap reports revealed that the human factors classifications were not informative for further analysis, and 70% of mishaps were attributed to unknown causes; only 23% were attributed to human factors. METHOD: Five diving fatality reports were examined using the consensual qualitative research (CQR) method to develop a taxonomy of six categories and 21 subcategories for classifying human errors in diving. In addition, 15 critical incident technique (CIT) interviews were conducted with U.S. Navy divers who had been involved in a diving accident or near miss and analyzed using the dive team error taxonomy. RESULTS: Overall, failures in situation awareness and leadership were the most common human errors made by the dive team. CONCLUSION: The dive team human error taxonomy could aid in accident investigation and in the training and evaluation of U.S. Navy divers. APPLICATION: The development of the dive team human error taxonomy has generated a number of considerations that researchers should take into account when developing, or adapting, an error taxonomy from one industry to another.     

Predicting risk as a function of risk factors

In previous work, we showed that risk factors have a significant negative effect on reliability (e.g., occurrence of failure). In this paper, we show that it is feasible to predict risk (i.e., the probability of risk factors being related to discrepancy reports occurring on the release of software). This is an important advancement over the previous research because discrepancy reports are available in the requirements phase—when the cost and labor required to correct faults is low—whereas failure data only becomes available in the test phase—when the cost and labor required for correcting faults is high. Although using historical failure data to drive traditional software reliability models would produce greater prediction accuracy, the opportunity to provide early prediction of reliability using risk factors outweighs this advantage.     

On perspective,expertise and differences in accident analyses: arguments for a multidisciplinary integrated approach

This study draws attention to the importance of the fact that accident analyses, their conclusions and recommendations for improvements differ depending on the purpose of the analysis and the analyst's basic professional training. An accident in a hospital in which three patients died during dialysis was used as an illustrative example. The result of the legal analysis is presented first. Using this perspective, the cause of the accident was judged to be errors made by the chief nurse who was the only one who was found guilty of the deaths of three people. She was given a conditional prison sentence. The action taken at the hospital was to remove her from her job. Later, engineering and psychology experts were asked about how they attributed responsibility for the accident. Finally, the Accident Evolution and Barrier Function (AEB) method was also used to analyse the accident. This method models an accident as an interaction between technical and human factors systems. The results illustrated how the conclusions concerning action to avoid an accident in the future differed widely following a legal analysis and an AEB analysis of the same accident. The data also showed that the responsibility for the accident was attributed to one single person in the legal analysis but chiefly to other agents by engineers and psychologists. Analysts with a basic training in engineering tended to find relatively more human factors errors than technical errors in the AEB analyses. The number of acceptable solutions created to avoid future errors were related to the analyst's basic professional training and/or motivation, but the basic AEB modelling of the accident evolution was relatively less dependent on the basic professional training of the analysts and/or motivation. It was argued that a legal framework can be inefficient or even contraproductive in promoting improvements to the safety of complex integrated systems. This is particularly true if the convicted people are low in the hierarchy of an organization and cannot affect its future routines. It is also argued here that experts with different basic professional training should perform accident analyses jointly. This is because only then will there be an insightful coverage of the interactions between different systems as, for example, human factors and technological systems making it possible to increase the safety of a complex integrated system.     

Identification of risk factors and countermeasures for slip, trip and fall accidents during the delivery of mail

Risk factors for slip, trip and fall accidents (STFA) during the delivery of mail were identified using a range of accident-centred and accident-independent methods. Key factors included slippery underfoot conditions, non-weather related environmental hazards (e.g., uneven paving, steps, inadequate lighting), poor slip resistance from footwear, unsafe working practices, management safety practices, and underlying organisational influences. Intervention measures were recommended that target STFA risks at three levels: slip resistance, exposure to hazardous conditions, and employee behaviour in the face of hazardous conditions. The use of a participative approach to intervention selection and design enabled allowance for the organisational context to be made.     

Fault-tree models of accident scenarios of RoPax vessels

Ro-Ro vessels for cargo and passengers (RoPax) are a relatively new concept that has proven to be popular in the Mediterranean region and is becoming more widespread in Northern Europe. Due to its design characteristics and amount of passengers, although less than a regular passenger liner, accidents with RoPax vessels have far reaching consequences both for economical and for human life. The objective of this paper is to identify hazards related to casualties of RoPax vessels. The terminal casualty events chosen are related to accident and incident statistics for this type of vessel. This paper focuses on the identification of the basic events that can lead to an accident and the performance requirements. The hazard identification is carried out as the first step of a Formal Safety Assessment (FSA) and the modelling of the relation between the relevant events is made using Fault Tree Analysis (FTA). The conclusions of this study are recommendations to the later steps of FSA rather than for decision making (Step 5 of FSA). These recommendations will be focused on the possible design shortcomings identified during the analysis by fault trees throughout cut sets. Also the role that human factors have is analysed through a sensitivity analysis where it is shown that their influence is higher for groundings and collisions where an increase of the initial probability leads to the change of almost 90% of the accident occurrence.     

Modelling of human alarm handling response times: a case study of the Ladbroke Grove rail accident in the UK

The purpose of the paper was to address the timeliness of the signaller's intervention in the Ladbroke Grove rail incident in the UK, as well as to consider the utility of human performance time modelling more generally. Human performance response time modelling is a critical area for Human Factors and Ergonomics research. This research applied two approaches to the same problem to see if they arrived at the same conclusion. The first modelling approach used the alarm initiated activity (AIA) model. This approach is useful for indicating general response times in emergency events, but it cannot comment in detail on any specific case. The second modelling approach employed a multi-modal critical path analysis (CPA) technique. The advantage of the latter approach is that it can be used to model a specific incident on the basis of the known factors from the accident inquiry. The results show that the AIA model produced an estimated response time of 17 s, whereas the CPA model produced an estimated response time of 19 s. This compares with the actual response time of the signaller of 18 s. The response time data from both approaches are concordant and suggest that the signaller's response time in the Ladbroke Grove rail accident was reasonable. This research has application to the modelling of human responses to emergency events in all domains. Rather than the forensic reconstruction approach used in this paper, the models could be used in a predictive manner to anticipate how long human operators of safety-critical systems might take to respond in emergency scenarios.     

Modelling of human alarm handling response times: a case study of the Ladbroke Grove rail accident in the UK

The purpose of the paper was to address the timeliness of the signaller's intervention in the Ladbroke Grove rail incident in the UK, as well as to consider the utility of human performance time modelling more generally. Human performance response time modelling is a critical area for Human Factors and Ergonomics research. This research applied two approaches to the same problem to see if they arrived at the same conclusion. The first modelling approach used the alarm initiated activity (AIA) model. This approach is useful for indicating general response times in emergency events, but it cannot comment in detail on any specific case. The second modelling approach employed a multi-modal critical path analysis (CPA) technique. The advantage of the latter approach is that it can be used to model a specific incident on the basis of the known factors from the accident inquiry. The results show that the AIA model produced an estimated response time of 17 s, whereas the CPA model produced an estimated response time of 19 s. This compares with the actual response time of the signaller of 18 s. The response time data from both approaches are concordant and suggest that the signaller's response time in the Ladbroke Grove rail accident was reasonable. This research has application to the modelling of human responses to emergency events in all domains. Rather than the forensic reconstruction approach used in this paper, the models could be used in a predictive manner to anticipate how long human operators of safety-critical systems might take to respond in emergency scenarios.     

Peer impressions in open source organizations: A survey

In virtual organizations, such as Open Source Software (OSS) communities, we expect that the impressions members have about each other play an important role in fostering effective collaboration. However, there is little empirical evidence about how peer impressions form and change in virtual organizations. This paper reports the results from a survey designed to understand the peer impression formation process among OSS participants in terms of perceived expertise, trustworthiness, productivity, experiences collaborating, and other factors that make collaboration easy or difficult. While the majority of survey respondents reported positive experiences, a non-trivial fraction had negative experiences. In particular, volunteer participants were more likely to report negative experiences than participants who were paid. The results showed that factors related to a person's project contribution (e.g., quality and understandability of committed codes, important design related decisions, and critical fixes made) were more important than factors related to work style or personal traits. Although OSS participants are very task focused, the respondents believed that meeting their peers in person is beneficial for forming peer impressions. Having an appropriate impression of one's OSS peers is crucial, but the impression formation process is complicated and different from the process in traditional organizations.     

Risk models for analyzing pilot-error at US airlines: a comparative safety study

Pilot-error is the major cause of accidents in airline operations. This paper proposes two risk models for analyzing pilot-error at US airlines. The focus is on whether airline-specific factors (the name of the airlines) are useful in predicting pilot-error. This is the first study that reliably models pilot-error accident and incident rates on an airline-by-airline basis. The results indicate that airline-specific factors generally were not useful predictors of pilot-error. Moreover, any predicted influence of an individual airline was substantially less than for the pilot-specific factors of age and experience. Thus, the paper illustrates the difficulty of trying to compare airlines on the basis of safety. By analyzing factors associated with pilot-error, policy-makers can more effectively manage and reduce the risk of airline accidents and incidents.     

Track maintenance train operators’ attitudes to job,organisation and management,and their correlation with accident/incident rate

The present paper reports the results of a questionnaire-based survey of night train operators attitudes toward management, operating procedures, and other organisational issues that potentially impact on safety. Responses were collected from all of the operators of track maintenance trains servicing the Japanese high-speed railway (Shinkansen). Two versions of the questionnaire, the TMAQ (Train Management Attitudes Questionnaire), were developed based on Helmreichs FMAQ (Flight MAQ) and its derivative, the SMAQ (Ship MAQ). The TMAQ and its progenitor seek to elicit respondents views of, and attitudes to, a range of safety related factors including morale, motivation, leadership and human relations in their organisation.To identify dimensions of safety culture as elicited through the TMAQ, a principal component analysis was applied to the questionnaire responses of the original TMAQ. The analysis yielded seven attitude factors, including morale and motivation. Of the seven factors, a close correlation was identified between the factor scores representing operators morale and motivation and the actual accident/incident rates for each of the five branches belonging to a single-track maintenance company. A branch that employed train operators having relatively higher morale and motivation exhibited a lower accident/incident rate. Furthermore, the very same correlation was also found for company based responses collected from all track maintenance companies working for the high-speed railway.In addition to the branch and company based comparisons for track maintenance train operators, we also compared attitude factors between different groups of operators (drivers and supervisors), and between two different periods surveyed in a two year interval. Finally, we examined differences in terms of attitude factors between track maintenance operators and seafarers surveyed by applying slightly different variants of the same generic form of questionnaire (Helmreichs SMAQ). Based on these survey results, we discuss potential risk factors for accidents of track maintenance trains and some implications for improving railway safety.     

Human factors data traceability and analysis in the European Community’s Major Accident Reporting System

This paper reports on human factors data traceability and analysis of the European Community’s Major Accident Reporting System (MARS). This is the main EU instrument to major accident data collection, analysis and dissemination for process industry according to the provisions of the Seveso II Directive. To date, the MARS database counts approximately 700 Seveso-type major events (November 2008). The MARS system is investigated in terms of human factors data and case studies exploring the relevance of human factors in accident events causation and identification issues. The human factors model and taxonomy as it is applied in the MARS system is reviewed. Criteria to enhance traceability and analysis about human related causes are then considered in detail. Finally, certain limitations affecting the system are pointed out. Findings are expected to favour future modelling and research efforts toward further MARS system improvements.     

Factors affecting employees' knowledge-sharing behaviour in the virtual organisation from the perspectives of well-being and organisational behaviour

Drawing on the concept of organisational behaviour, this research augments the concepts of social capital theory and organisational culture with one pioneering precursor and mediator, the sense of well-being, to develop an integrative understanding of the factors affecting individuals' knowledge-sharing behaviour within the more complex context of the virtual organisation of Taiwanese Non-governmental Organisations (NGOs). A field survey of 131 employees from the selected virtual organisation was analysed using Structural Equation Modeling (SEM) to examine the outcomes empirically. Our research offers a persuasive body of evidence supporting the notion that increasing employees' sense of well-being can successfully form a bridge that can connect social capital tendency, organisational culture and employees' knowledge-sharing behaviour. Surprisingly, and contrary to common belief, the integrated model shows that social capital tendency seems to play a more important role than organisational culture in affecting employees' sense of well-being within the virtual organisation in a Chinese cultural context. Consequently, this research reveals the subtle interplay of employees' sense of well-being, social capital tendency, organisational culture and knowledge-sharing behaviour, while the in-depth analysis provides strong support for knowledge management research and practice.     

Using Reason’s Model of Organisational Accidents in Formalising Accident Reports

This paper proposes a method for analysing what are called organisational accidents. The first step of the method involves using Reason’s model of organisational failures. This provides heuristic guidance in identifying both the active and latent conditions that lead to major failures. The second step involves applying formal methods to support a detailed analysis of each latent and active condition. The method is demonstrated on a case study: the railway accident at Watford Junction in the United Kingdom. Analysis of the formal model helps to identify organisational factors that might have prevented the accident. It also helps to identify weaknesses in the report itself. In particular we argue that a signalling standard was misunderstood, the consequences of which could lead to another serious accident.     

Assessing the benefits from e-business transformation through effective enterprise management

This paper reports on research carried out in 1999–2001 on the use of e-business applications in enterprise resource planning (ERP)-based organisations. Multiple structured interviews were used to collect data on 11 established organisations from a diverse range of industries. The findings are analysed according to the level of sophistication of e-business models and their transformational impact on the organisation. Early adopters of e-business show a trend towards cost reductions and administrative efficiencies from e-procurement and self-service applications used by customers and employees. More mature users focus on strategic advantage and generate this through an evolutionary model of organisational change. Two complex case studies of e-business integration with global suppliers and their corporate customers are analysed to identify specific stages of benefits accrual through the e-business transformation process. Collectively, the set of case studies is used to demonstrate the increased benefits derived from an e-business architecture based on a network of ERP-enabled organisations.     

Maintaining production quality in advanced mechanical engineering organizations: The compensatory function of the CNC‐machine operator

This article reports a study that was carried out in 6 German advanced mechanical engineering organizations. The research investigated the wider role of the CNC machine operator against the background of an increased focus on quality management issues and lean production. Fifty‐one interviews were carried out with individuals from different professional groups. The results showed that operators had an important function in compensating for variations in the manufacturing process. The choice of operator actions was influenced by several production‐related factors (e.g., tolerance limit). Furthermore, we identified several organizational boundaries that indicated a considerable potential for efficiency gains if collaboration across these boundaries could be improved. © 2000 John Wiley & Sons, Inc.