Journal of Safety Research最新文献

{"title":"Teen perceptions of parental monitoring and its impact on their risky road behavior: An analysis of the National Youth Risk Behavior Survey","authors":"Amir Ghanbari ,&nbsp;Matison Howard ,&nbsp;Joseph Cavanaugh ,&nbsp;Cara Hamann","doi":"10.1016/j.jsr.2025.03.007","DOIUrl":"10.1016/j.jsr.2025.03.007","url":null,"abstract":"<div><div><em>Introduction:</em> Risky road behaviors (RRBs), including driving-related (e.g., texting while driving, driving under the influence) and passenger-related (e.g., not wearing seat belts, riding with a drunk driver) behaviors, contribute significantly to injury and death among adolescents. This study aims to analyze how perceived parental monitoring by teens impacts their passenger- and driver- related risky road behaviors. <em>Methods:</em> Data from the 2021 National Youth Risk Behavior Survey (NYRBS) were analyzed to examine the association between teens’ perception of parental monitoring and engagement in driving-related and passenger-related risky road behaviors. Logistic regression models were used to estimate adjusted odds ratios (AOR). <em>Results:</em> A strong perception of parental monitoring consistently demonstrated a protective effect against driving-related risky behaviors (AOR = 0.63, 95% CI: 0.53–0.74) and passenger-related risky behaviors (AOR = 0.62, 95% CI: 0.53–0.72) when compared to those with lower perceptions. Teens sleeping in public places had notably higher odds of driving-related risky behaviors (AOR = 2.99, 95% CI: 2.01–4.46) compared to those sleeping at home. Males were less likely to engage in passenger-related risky behaviors (AOR = 0.91, 95% CI: 0.86–0.97) but more likely to engage in driving-related risky behaviors (AOR = 1.28, 95% CI: 1.12–1.47). <em>Conclusions:</em> Perceived parental monitoring was associated with reduced engagement in both passenger- and driving-related risky road behaviors among teen drivers and passengers. Safety interventions aimed at reducing risky road behaviors among teens should include strategies for strengthening parental involvement, emphasizing supervision and communication. <em>Practical Applications:</em> Our findings suggest that parental monitoring could play a significant role in reducing teens’ engagement in risky road behaviors. Safety programs should empower parents in the use of effective parental monitoring strategies, such as active supervision and better communication.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 342-347"},"PeriodicalIF":3.9,"publicationDate":"2025-04-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143800387","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Improving the usability of large emergency 911 data reporting systems: A machine learning case study using emergency incident descriptions","authors":"N. Katherine Yoon ,&nbsp;Tyler D. Quinn ,&nbsp;Alexa Furek ,&nbsp;Nora Y. Payne ,&nbsp;Emily J. Haas","doi":"10.1016/j.jsr.2025.04.001","DOIUrl":"10.1016/j.jsr.2025.04.001","url":null,"abstract":"<div><div><em>Introduction:</em> Emergency 9-1-1 incident data are recorded voluntarily within fire-department-specific computer-aided dispatch systems. The National Fire Incident Reporting System serves as a repository for these data, but inconsistency and variability in reporting practices across departments often lead to challenges in data quality and utility. This study aims to enhance emergency incident categorization and explore the feasibility of an automated system using free-text incident data from the National Fire Operations Reporting System (NFORS). <em>Method:</em> Researchers extracted and standardized 3,564 unique 9–1-1 incident descriptions from six fire departments using NFORS data, including narrative fields from emergency reports. The data were preprocessed using natural language processing (NLP) techniques, such as tokenization, stop word removal, and feature extraction (e.g., TF-IDF and n-grams). These features were used to train and evaluate Machine Learning (ML) models, including Naïve Bayes, Random Forest, and Support Vector Machine, to classify incidents into nine categories. The NLP techniques prepared the text data for the ML models, which performed the classification and assessed the automated system’s performance. <em>Results:</em> The study demonstrated significant improvements in incident categorization accuracy using the NLP and ML approach. Unigram models achieved 93% accuracy when applied to 3,564 unique incident descriptions. This performance was evaluated by comparing the automated classifications to manually assigned categories, which served as the reference. Mis-categorizations primarily occurred with “Emergency Medical Services (EMS).” <em>Conclusions:</em> Standardized and consistent incident categorization is vital for informed decision-making, efficient resource allocation, and effective emergency response. Our findings suggest that adopting a robust categorization system, such as the nine-category model using NLP and ML, can improve categorization accuracy and enhance data quality and utility for decision-making. <em>Practical Applications:</em> Public safety agencies can leverage these insights to modernize data systems, strengthen occupational surveillance, and create more resilient and sustainable public safety data systems.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 335-341"},"PeriodicalIF":3.9,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143785140","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and validation of a video-based hazard prediction test for e-bike riders in China","authors":"Long Sun,&nbsp;Meiqi Zhang,&nbsp;Yunong Ma","doi":"10.1016/j.jsr.2025.03.008","DOIUrl":"10.1016/j.jsr.2025.03.008","url":null,"abstract":"<div><div><em>Purpose:</em> The present study aimed to develop a hazard prediction test for e-bike riders using the hazard prediction paradigm. The validity of the test was examined by comparing its results with retrospective traffic accidents in the preceding three years and with prospective traffic accidents in the 12 months following the survey. <em>Method:</em> Study 1 employed a single-factor experimental design and presented 28 video clips to 60 experienced riders and 60 novice riders. Study 2 examined the known group validity of the newly developed test by comparing the differences in the test scores of e-bike riders who had been involved in accidents in the preceding three years (n = 53) and those who had not (n = 59) and examined the predictive validity of the test using negative binomial regression analysis by exploring whether the test score could predict accident involvement in the following 12 months. <em>Results:</em> The results of Study 1 show that the newly developed hazard prediction test contains 20 videos with satisfactory reliability (Cronbach’s alpha = 0.78), and experienced e-bike riders scored significantly higher than novices did on the test. The results of Study 2 show that e-bike riders involved in accidents in the preceding three years had lower test scores than did those not involved in accidents. More importantly, the test scores effectively predicted the crash risk of e-bike riders in the following 12 months, suggesting that the predictive validity of the test is satisfactory. <em>Conclusions:</em> The hazard prediction test for e-bike riders has acceptable reliability and validity and can be used as a reliable tool to assess the hazard perception of e-bike riders.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 327-334"},"PeriodicalIF":3.9,"publicationDate":"2025-04-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143767655","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Special Report from the CDC: Racial/ethnic differences in motor-vehicle crash deaths, United States, 2019–2022","authors":"Sarah Matthews,&nbsp;Dionne Williams,&nbsp;Amy Schumacher,&nbsp;Avital Wulz,&nbsp;Jennifer Monda,&nbsp;Jason Hymer,&nbsp;Holly Billie,&nbsp;Rebecca B. Naumann,&nbsp;Bethany A. West","doi":"10.1016/j.jsr.2025.03.005","DOIUrl":"10.1016/j.jsr.2025.03.005","url":null,"abstract":"<div><div><em>Introduction:</em> Motor-vehicle crash (MVC) deaths increased by a record 10% from 2020 to 2021 in the United States and disproportionately impacted persons of certain racial/ethnic groups. <em>Methods:</em> Mortality data from the National Vital Statistics System was used to describe MVC death rate trends during 2019–2022 by six racial/ethnic groups: non-Hispanic (NH) American Indian or Alaska Native (AIAN), NH Asian, NH Black, NH Native Hawaiian or Other Pacific Islander (NHOPI), NH White, and Hispanic. Age-adjusted death rates per 100,000 population, 95% confidence intervals (CIs), and annual percent change in rates were calculated. <em>Results:</em> Overall MVC death rates increased during 2019–2022, and rates were highest among NH AIAN and NH Black persons across all years. During 2019–2020, death rates increased the most among NH Black persons (+26.0%). During 2020–2021, rates increased among all racial/ethnic groups, with the greatest increase among NH NHOPI persons (+66.7%) and NH AIAN persons (+27.8%). <em>Conclusions:</em> These findings highlight stark differences by racial/ethnic group in MVC death rates and changes in these rates. Between 2019 and 2022, NH AIAN, NH Black, and NH NHOPI populations experienced the largest increases in MVC death rates, although there was large variation in rates and trends. Widespread adoption of a comprehensive suite of prevention strategies, such as the Safe System approach, while targeting subpopulations with the greatest burden of MVC deaths could reduce these differences and the overall burden of MVCs. <em>Practical Applications:</em> These findings show which subpopulations could experience the greatest impacts from transportation safety investments in reducing overall MVC death rates in the United States.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 441-446"},"PeriodicalIF":3.9,"publicationDate":"2025-03-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144223506","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Organizational climate and safety climate are nearly inseparable in shared living and working spaces","authors":"Adam T. Biggs ,&nbsp;Jason T. Jameson ,&nbsp;Todd R. Seech ,&nbsp;Peter G. Roma ,&nbsp;Rachel R. Markwald ,&nbsp;Dale W. Russell","doi":"10.1016/j.jsr.2025.02.026","DOIUrl":"10.1016/j.jsr.2025.02.026","url":null,"abstract":"<div><div><em>Introduction:</em> Safety climate represents a shared collection of attitudes, perceptions, and best practices concerning safety behaviors within an organization. Among the many antecedents that potentiate good safety climate conditions, a reasonably robust predictor has been organizational climate. This construct represents a more comprehensive view of attitudes beyond just the safety-related behaviors that make up safety climate. Although this comparison has been explored across different industrial settings, what happens to the relationship between organizational climate and safety climate when people live and work within the same environment? Several industrial settings would satisfy this criterion, including mining, oil drilling, nascent spacefaring missions, and maritime operations, with the latter representing a unique context where people of various job types must work and live together for extended periods of time. <em>Method:</em> The current study thus explored the relationship between organizational climate and safety climate and safety issues using a large sample of U.S. Navy personnel (N &gt; 11,000) from more than 100 ships at sea. <em>Results:</em> Structural equation modeling revealed a strong relationship between organizational climate and safety climate (<em>β</em> = 0.92) that is consistent across gender, age, and paygrade. However, the relationship between organizational and safety climates and safety issues varied across organizational tiers, suggesting that climate strength may vary across hierarchal levels. <em>Conclusions:</em> These findings suggest that organizational climate and safety climate become nearly inseparable when personnel must live and work within the same environment, although climate strength may represent a separate issue when examining impact on actual safety outcomes. <em>Practical Applications:</em> Practical applications are discussed, including increased monitoring of organizational climate when assessing safety behaviors in high-risk and highly enmeshed work environments.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 405-412"},"PeriodicalIF":3.9,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144223503","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Sex- and age-specific differences in product-related traumatic brain injury morbidity in the United States, 2004–2023","authors":"Wangxin Xiao ,&nbsp;David C. Schwebel ,&nbsp;Lei Yang ,&nbsp;Min Zhao ,&nbsp;Shuying Zhao ,&nbsp;Peixia Cheng ,&nbsp;Guoqing Hu","doi":"10.1016/j.jsr.2025.03.004","DOIUrl":"10.1016/j.jsr.2025.03.004","url":null,"abstract":"<div><div><em>Introduction:</em> To examine trends in product-related traumatic brain injury morbidity by sex and age group in the Unites States between 2004 and 2023. <em>Method:</em> Product-related injury morbidity data were extracted from the National Electronic Injury Surveillance System (NEISS). Eight types of products, five injury locations, two sexes, and five age groups were categorized. Joinpoint regression models were performed to detect time periods showing significant product-related traumatic brain injury morbidity changes between 2004 and 2023. Average annual percent changes (AAPCs) in morbidity rates and 95% confidence intervals (95% CIs) quantified significant morbidity changes. <em>Results:</em> The age-standardized product-related traumatic brain injury morbidity rate more than doubled in the United States between 2004 and 2023, rising from 236.6 to 574.8 per 100,000 persons (AAPC = 5.1%, 95% CI: 3.7%, 6.5%), with notable fluctuation during the COVID pandemic in 2019–2021. Under-15 children and old adults ≥ 65 years had the highest product-related traumatic brain injury morbidity rates between 2004 and 2023. Floors/flooring materials and stairs/steps were the most common products causing traumatic brain injuries, accounting for 31.7% of injuries, and homes were the most frequent occurring location, accounting for 59.3% of total product-related injury morbidity. Morbidity rates and spectrums by type of product and by occurring location varied greatly across sex and age groups. <em>Conclusions:</em> Product-related traumatic brain injury morbidity rates increased in the United States between 2004 and 2023, with some morbidity fluctuations in 2019–2021, likely reflecting the effect of the COVID-19 pandemic. <em>Practical Applications:</em> Further research and prevention efforts are recommended to interpret the observed morbidity changes, curb recent product-related traumatic brain injury morbidity increases and reduce morbidity disparities across sex and age groups.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 318-326"},"PeriodicalIF":3.9,"publicationDate":"2025-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143697875","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Impact of an in-vehicle audio prompt on multistage lane-changing behavior and gaze patterns in work zone","authors":"Ke Duan ,&nbsp;Xiaomeng Li ,&nbsp;Xuedong Yan ,&nbsp;Yan Huang ,&nbsp;Junyu Hang ,&nbsp;Qingwan Xue","doi":"10.1016/j.jsr.2025.02.024","DOIUrl":"10.1016/j.jsr.2025.02.024","url":null,"abstract":"<div><div><em>Introduction:</em> Lane changing poses a great threat to traffic safety in the work zone areas. This study aims to conduct an in-depth investigation of drivers’ lane-changing behavior and to propose and verify the efficacy of an intelligent in-vehicle audio prompt through a driving simulator experiment. <em>Method:</em> Forty-two participants, comprising 22 males (12 professional drivers and 10 normal drivers) and 20 females (10 professional drivers and 10 normal drivers), took part in the study. Three traffic density levels (20pcu/km, 40pcu/km, and 60pcu/km) and five locations (baseline, 250m, 500m, 750m, and 1000m) to release the audio prompt information were set to collect eye movement and driving behavior data. A multistage lane-changing process including the perception stage, preparation stage, and action stage was developed, and three types of variables, namely experiment variable, behavioral variable, and risk assessment variable, were defined. <em>Results:</em> The results indicated that the audio prompt set at 500m, 750m, and 1000m could noticeably improve lane-changing behavior (i.e., triggering lane changes more in advance and providing drivers with more preparation time). However, the audio prompt set at 250m upstream of the work zone showed limited efficacy, which is not recommended in practical application. Male drivers and professional drivers performed better during lane changing, and they had a more active observation of the surrounding environment and tended to adopt safer vehicle operations than their counterparts. <em>Practical Applications:</em> A high traffic density could increase the collision risks even if drivers had made sufficient preparations, while in low traffic density conditions, drivers who failed to change lanes tended to apply hard braking when they approached the work zone.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 298-317"},"PeriodicalIF":3.9,"publicationDate":"2025-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143686246","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development of roundabout crash-specific safety performance functions in Kentucky using different techniques","authors":"Tathagatha Khan,&nbsp;Bharat Kumar Pathivada,&nbsp;Arunabha Banerjee,&nbsp;Kirolos Haleem","doi":"10.1016/j.jsr.2025.03.001","DOIUrl":"10.1016/j.jsr.2025.03.001","url":null,"abstract":"<div><div><em>Introduction</em>: Roundabouts are one of the innovative intersection designs that can reduce the number of fatal and serious injury crashes compared to traditional intersections. Nevertheless, relatively limited studies are available that have developed roundabout crash-specific safety performance functions (SPFs) using different modeling techniques. This study develops roundabout crash-specific SPFs using crashes in the state of Kentucky over a four-year period (2019–2022). <em>Method:</em> <!-->Crash data exhibited over-dispersion; hence, negative binomial (NB) and heterogenous negative binomial (HTNB), as well as Conway-Maxwell-Poisson (CMP) modeling approaches with fixed and varying dispersion parameters (while accounting for unobserved heterogeneity) were investigated. The performance of the fitted models was evaluated using different goodness-of-fit (GOF) statistics, e.g., Akaike information criterion “AIC,” Bayesian information criterion “BIC,” and McFadden pseudo R². <em>Results:<!--> </em>The GOF measures showed that the heterogeneous CMP (or HTCMP) model was the best-fit model for developing roundabout SPFs. The results of the HTCMP model showed that annual average daily traffic (AADT) on major and minor approaches to the roundabout and inscribed circle diameter (ICD) were significantly associated with increased crash frequency at roundabouts. Conversely, single circulatory lanes, presence of two lanes along the major roundabout approach, and wider circulatory roadway width (greater than 16 feet) were significantly associated with reduced crashes at roundabouts. <em>Practical Applications:</em> <!-->Based on the study results, safety countermeasures were proposed to enhance safety at roundabouts. One example is ensuring roundabout circulatory roadway widths greater than 16 feet to provide adequate room for driving error rectification while navigating roundabouts.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 282-291"},"PeriodicalIF":3.9,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143621173","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The human cost of allowing unhelmeted motorcycling in the United States","authors":"Eric R. Teoh","doi":"10.1016/j.jsr.2025.03.002","DOIUrl":"10.1016/j.jsr.2025.03.002","url":null,"abstract":"<div><div><em>Introduction:</em> This study’s objective was to estimate the number of motorcyclist fatalities attributable to laws that allow unhelmeted riding in the United States since 1976. <em>Method:</em> Counts of helmeted and unhelmeted motorcyclist fatalities were used to estimate population-level helmet use under all-rider helmet laws and in the absence of such laws. The number of lives that could have been saved if helmet use in states that allowed unhelmeted riding was equal to helmet use in states with all-rider helmet laws was estimated for each year and summed over the study years. <em>Results:</em> If all states had all-rider helmet laws throughout the 1976–2022 study period, 22,058 fewer motorcyclists would have died in crashes. This represents 11% of all motorcyclist fatalities during these years. The number of motorcyclists killed in 2022 would have been 10% lower. Currently, 17 states and the District of Columbia have an all-rider helmet law in place. <em>Conclusion:</em> Requiring helmets for all motorcyclists is a straightforward rule of the road that has the potential to reduce annual motorcyclist fatalities, which are at record-high levels of over 6,000 per year, by 10%. All-rider helmet laws are a fundamental component of a Safe System for motorcycling. <em>Practical Applications:</em> States should consider the human cost of not having all-rider helmet laws and use this tool to reduce the number of riders killed in crashes.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 292-297"},"PeriodicalIF":3.9,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143629089","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Operational safety risk modeling in a naval organization","authors":"Dale W. Russell ,&nbsp;Ryan Lance ,&nbsp;Patrick J. Rosopa","doi":"10.1016/j.jsr.2025.02.025","DOIUrl":"10.1016/j.jsr.2025.02.025","url":null,"abstract":"<div><div><em>Introduction:</em> Following numerous mishaps and near-misses, the U.S. Naval Surface Force established the Operational Surface Risk Indicators (OSRI) project to explore a robust proactive risk analysis and reduction capability. The OSRI model leverages multisource enterprise-wide data to forecast potential risks for individual warships to help leaders make informed operational decisions and ultimately improve safety outcomes. <em>Method:</em> Machine learning was used to predict risk scores based on various input features that describe a ship’s current state, including crew training, staffing levels, experience, and turnover. Machine learning techniques like stacking and bagging were employed in novel ways to enhance model interpretability and fairness. To make the model’s output more tangible, predicted risk scores were converted into calibrated probabilities of mishaps; additionally, nearest neighbor techniques were integrated to provide insights on how current high-risk ships may be similar to past ships that experienced mishaps. <em>Results:</em> The model highlights risk indicators as potential areas for improvement and incorporates additional methods to derive insights from risk modeling. <em>Practical applications:</em> From an operational perspective, the OSRI model aims to enhance leaders’ intuition about risk by identifying indicators that are drivers of risk.</div></div>","PeriodicalId":48224,"journal":{"name":"Journal of Safety Research","volume":"93 ","pages":"Pages 274-281"},"PeriodicalIF":3.9,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143610510","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}