著書

安全のためのシステム（対策や教育を含む）の実効性を高めるためには適切なマネジメントが必要です。形骸化しやすい安全管理の仕組みを的確にマネジメントすることにより、企業のサービスや製品が利用者だけでなく地域社会にまで安心をもたらすことができます。的確な安全マネジメントは企業のレピュテーション（評判）をも向上させていきます。　 　本書ではヒューマンエラー対応を含む安全対策を企業価値向上のための戦術と捉え、経営戦略との整合を図りながら、事業者のレピュテーションを高める安全活動の体制を構築していくプロセスについて解説します。

In recent years, safety management has been increasingly strengthened, and employees are now expected to voluntarily assess safety awareness and organizational culture. However, the evaluation of motivation for safety activities is primarily conducted through surveys and interpreting the results and identifying issues often rely on expert assessments. This approach is costly, making detailed analysis at the departmental level difficult. This study aims to develop a system that automatically analyzes the results of safety awareness surveys from multiple perspectives and suggests challenges and solutions for safety management. Using data collected from over 70 businesses, totaling approximately 200,000 individuals since 2007, we modeled the relationships using variance-covariance structure analysis and formulated correlations with expert assessments through machine learning (deep learning). As a result, we developed an application that automatically identifies current challenges in safety activities and provides potential solutions. In summer 2024, the system was implemented in a business with approximately 15,000 employees to evaluate the validity of management advice for on-site operations. Moving forward, we plan to expand the system to industries such as railways, manufacturing, IT services, and healthcare to enhance its practical application.

95 % of occupational accidents are said to be caused by unsafe human behavior. In order to prevent occupational accidents, various types of human resource education are implemented in each company, but the current situation is that many of them are aimed only at preventing serious accidents in steady state. One of the reasons for this is that while many companies have substantial training programs that use concrete examples, they still do not have enough training to develop the ability to apply these concrete examples to new cases that they actually face. The ability to accurately summarize concepts from concrete examples and grasp things as a whole is called conceptual skill.Organizations with on-site leaders with high conceptual skills not only have a low occupational accident rate but can effectively organize teams to achieve organizational goals. Therefore, in this study, we will propose a failure event virtual experience-based learning method that enables the development of conceptual skills instead of on-site experience, and measure its effect, by investigating various conventional educational methods. Many text-based approaches, such as interviews and questionnaires, have been conducted in order to acquire workers’ conceptual skills, but there are still few examples of behavior-based study. Therefore, in this study, we propose a conceptual skill acquisition support method through analyzing conceptual skills from the behavior of subjects.

To improve the quality of safety activities, many companies are aiming to increase employees’ understanding of safety activities and motivate them to participate in them. The main method to measure the willingness to participate is through questionnaires, but it takes time and effort. Therefore, we focused on information on “Awareness” events reported by on-site employees in daily work report. We call this information “Information on Preparing for the Unexpected”. The purpose of this study was to develop a method to evaluate employees’ understanding of and willingness to participate in safety activities using “Information on Preparing for the Unexpected”. Through experiments, trouble-related data were collected at several workplaces, and text mining was conducted. We worked on the construction and validation of the system.

Currently, many companies provide human error response training. The objective of this training is to gain knowledge about human error, participate in safety activities on their own, and make use of this knowledge in actual workplaces. However, many companies do not properly measure the effectiveness of education. Therefore, this study examined the creation of a model that can evaluate the effectiveness of human error response education. We created a new questionnaire tool by referring to engagement surveys currently conducted in various companies and questionnaires that measure the personality of individual characteristics. We then conducted a questionnaire at an IT company that conducts human error response education and attempts to measure its effectiveness in terms of exercise scores. Multiple regression analysis was conducted based on the results of the questionnaire and the exercise scores, and a model was created to enable measurement of the effectiveness of the education. In this study, we were able to find clues for creating an evaluation index to measure the effectiveness of human error response education. However, the measurement of effectiveness was unclear in some cases, and issues remained regarding the accuracy of the measurement. Once this model is established, it is expected that companies that have not yet been able to measure the effectiveness of their human error response training will be able to do so by using a simple questionnaire. Based on the results of this study, we plan to further expand the data and create an evaluation index that will enable more accurate measurement of the effectiveness of education.

Preventive measures against various human errors are being taken based on information on near-miss events. However, the process from collecting near-miss events to analysing them and planning countermeasures is labour-intensive. Focusing only on high-risk near-miss events can reduce labour, but many of the collected nearmiss events will not be used. To solve these problems, we believe that a “near-miss event analysis support system” consisting of the following tools will be useful. (1st Tool) A tool to support the analysis of near-miss events, especially factor analysis (automatically extracts factors and identifies human error type. Analysts can add or modify.) (2nd Tool) A tool to support risk assessment of near-miss events (calculates the possibility of human error occurring in the target work on a 5-point scale. Analysts input the 5-point scale based on the expected extent of damage when it occurs. These two values are multiplied and the risk is evaluated on a 5-point scale. (3rd Tool) A tool to present countermeasures for near-miss events that are judged to be high risk (automatically presents appropriate human error countermeasure policies and three specific candidate measures based on human error type. Analysts select countermeasures based on the presented countermeasure policies) (4th Tool) Nearmiss event occurrence trend analysis tool (Automatically performs statistical analysis of the causes of near-miss events that occurred during a period set by the analyst. Also performs categorization analysis based on the work site, work time period, and SRK level of the work.) (5th Tool) Near-miss event management tool (Connects the near-miss event input tool with the above four tools, stores all data such as evaluation results in the cloud, and supports horizontal deployment within the company. Based on the evaluation results, the urgency is evaluated in four stages (Level 0∼3), and if it is Level 2 or above, a function is added to automatically contact related departments from the system.) We implemented these five tools based on AI technology and built a near-miss event analysis support system. This system is currently being test-operated by safety personnel from several companies, and although we are still in the process of collecting operational issues, it has been confirmed that it has the expected effects.

A major issue in the analysis of in-hospital incidents is the lack of skill and experience of the analyst. When an analyst with little experience analyzes an incident, they tend to only extract superficial characteristics and list only ineffective measures such as minor work improvements and thorough confirmation. To solve this issue, it is essential to educate the analyst. However, the higher the safety of the organization, the less experience employees have with accidents and serious incidents. Compensating for this experience with only regular education and virtual experiences is too laborious and takes a lot of time for the analyst to grow. Therefore, in this study, we examined strategies to enhance the awareness of analysts through the routine analysis of incident reports in medical settings. Specifically, this involves the development of the Medical Risk-Managers’ Awareness Enhancement System in Medical Incident Analysis.

Near-miss analysis is essential for identifying factors contributing to human errors and developing preventive measures. However, conventional text mining methods primarily extract direct causes, such as “lack of attention” or “insufficient verification,” often overlooking broader background factors embedded in work environments. This study explores the use of Large Language Models (LLMs) to enhance factor analysis by capturing a more comprehensive range of underlying causes. Using Llama3-ELYZA-JP-8B, we incorporated 87 predefined background factors based on established frameworks, including PSF lists and the m-SHEL model. The developed factor analysis support system was applied to actual near-miss reports, and its effectiveness was evaluated by comparing the number and diversity of extracted factors before and after implementation. Results showed that LLM-based analysis significantly increased factor extraction and enhanced the identification of diverse causes. Additionally, factor aggregation and visualization improved the interpretation of trends over time. Despite these advantages, challenges remain, particularly regarding biases in data, factor extraction, and decision-making. Future research should focus on managing these biases through data diversity, optimized extraction balance, and improved transparency in analysis. By addressing these issues, a more reliable and practical near-miss factor analysis support system can be developed, contributing to improved workplace safety and more effective error prevention strategies.

In medical facilities and general hospitals, a variety of information regarding incidents in patient care is collected and analysed. However, due to the large differences in knowledge and experience of on-site risk managers, previous studies have shown several problems with the data collected. Underlying these problems is the analyst’s limited knowledge of human factors, IT, and management. However, it is very difficult to give more time and cost for safety training to on-site risk managers. Therefore, in this study, we decided to organize the results of past research on human error factor analysis from the perspective of on-site management. As a result, we were able to obtain a set of elements [elements for activation of on-site activities] for improving the ability of on-site risk managers to recognize, the willingness of on-site risk managers to participate in medical safety activities, and the management level of on-site risk managers. Based on the “Elements for Activation of On-Site Activities”, we developed a prototype of an incident reporting support system.

In factories such as aircraft manufacturing, where the number of productions is small and the defective rate must be reduced to zero, a large number of human error prevention measures are taken. “CRM” in the airline industry is a specific example of ones. However, the burden on workers due to too many measures has exceeded the limit, and there is an urgent need to optimize the management of error prevention measures as a whole. To achieve this, (1)Detailed collection of human factor information on problematic events such as nonconformity events, (2) Collection of human factor information during normal times of the target work, (3) Structural analysis of human factor, (4) Proposal of guidelines for human error prevention measures based on the analysis results and presentation of multiple specific measures. Of these, (3) and (4) have already been developed in our laboratory and are at the practical stage. Structural analysis methods for human factors include the Swiss cheese model, m-SHELL, PSF list and variation tree. It has been confirmed that (1) can be resolved by measuring human factor management courses for work team leaders through e-learning and factor analysis training incorporating active learning for six months. The realization of safety among business operators has also changed from the traditional “Safety-I” to “Safety-II,” and the demand for (2) is increasing, but method (2), which seeks to discover issues when no problems have occurred at all, cannot be addressed with methods such as traditional near-miss event analysis. Therefore, in this study, the image of collected information was changed to “hints that lead to good work” and methods of collecting human factor information from on-site conversations were examined, centering on stimulating constructive communication on-site.

For the proposal of adequate safety measures in industrial plants, it is very important to conduct an accurate risk assessment. A conventional risk assessment aims to eliminate hazards in the hardware aspects, which have a large impact. In contrast, human factors, which has a smaller impact, is not analyzed in detail, and is evaluated only qualitatively or by using a constant value. However, as the number of hazards is decreasing, it is necessary to analyze human factors as well. The purpose of this study is to support the enhancement of incident-level risk assessment. We examine the assessment method of human error probability that can be conducted by those with limited knowledge of human factors.

Natural disasters pose various challenges to the national land. While various technological innovations have contributed to overcoming these difficulties, there are still many problems that need to be solved to play a preventive role. A framework is needed to examine social design proposals that promote the production of technology. However, in order to implement such an idea in society, a new business model based on non-monetary value and reputation must be developed to address various social issues, including not only disaster prevention but also maintenance and management of social infrastructure and green infrastructure, including agriculture. This research is therefore focused on the concept of adaptive finance, which is a practical approach to finance for adaptation. Therefore, the purpose of this study was to organize basic data and visualize the characteristics of new technologies in order to put the concept of adaptive finance into practice. We examined ways to visualize the effects of new technologies based on a database of new technologies (domestic open data) that have been developed for the maintenance and management of social infrastructure and for disaster prevention and mitigation. In addition, we developed an index to quantitatively evaluate the degree of damage reduction. By doing so, we aimed to clarify how the introduction of new technologies contributes to disaster prevention and mitigation, and to promote their application to actual disaster prevention policies. For this visualization, we used generative AI to create an index based on various technological data. The results of this research will support policy makers, local governments, and private companies in their decisions to introduce disaster prevention technologies and contribute to minimizing disaster damage in the future.

One of the important issues in the field of safety management is effective human error prevention education for on-site staff. Currently, many sites not only provide basic human error response training such as confirmation and thorough implementation of basic actions, but also various education methods such as work improvement that takes human factors into consideration, active follow-up with team members, and communication methods that lead to accurate reporting, contact, and consultation. However, the effectiveness of such human error prevention education has only been measured qualitatively. Therefore, this study focused on the engagement of workers and examined a method to multifacetedly evaluate each trainee’s attitude toward safety activities before and after the course, including the individual characteristics of the workers (personality such as personality analysis). A questionnaire was designed with work engagement, personal engagement, burnout, employee engagement, psychological safety, personality (Big Five theory), and attributes (job type, position, years of experience) as basic parameters, and a model was created to evaluate the following four main factors based on the answers. The four main factors are: (i) loyalty to work, (ii) desire for growth, (iii) desire to contribute to the company, and (iv) sense of happiness (well-being). These indicators were evaluated on a 10-point scale. The effectiveness of the method of measuring the effectiveness of on-site human error prevention training based on the employee engagement indices obtained in this study was verified through surveys at several hospitals in Japan. Since the measurement of effectiveness may be unclear in some cases, we are still working to expand the accuracy of the method by expanding the data.

Patients’ falls and fall-related incidents account for approximately 24% of all medical accidents, emphasizing the need for enhanced preventive measures. Traditional fall risk assessments do not consider nursing systems and impose a significant operational burden. In this study, we developed a simulation program that replicates hospital nursing systems and fall duration, analyzing the relationship between the number of nurses and fall duration. The simulation results showed that as the number of nurses increased, fall duration decreased. However, the effectiveness plateaued when the number exceeded eight nurses. Additionally, the distribution of fall duration followed a log-normal distribution, allowing it to be formulated as a costperformance optimization problem. As a result, the optimal number of nurses per floor was suggested to be between six and eight. Future research will incorporate nurse patrol routes and skill differences into the simulation to improve its applicability to real hospital environments. Collaborating with medical safety managers, we aim to develop and implement highly practical fall prevention measures.

In heavy manufacturing, where the production volume is low and the defect rate must be zero, extensive human error countermeasures have been implemented. However, the excessive number of countermeasures has placed a heavy burden on workers, necessitating the optimization of error management strategies. This study focuses on gathering human factor information during routine tasks by facilitating constructive communication in the workplace. A system was designed to evaluate this information multidimensionally, using positive words as indicators of constructive communication.