Search

Abstract: Navigation systems are very useful tools because they display a user''s location and guide them to a destination using graphics, text and voice information. Recent work has revealed that millions of consumers received driving directions using their cell phone or PDA. This present work aimed to explore whether the efficiency to destination and driver behavior were distinguishable when using a portable navigation system compared to a paper map. Thirty-two subjects were paid to participate in this research, with field experiments being carried out in both urban and rural environments. A smart phone was adopted as the portable navigation system in the study. The results revealed that the drivers performed better when using a portable navigation system compared to those using a paper map, in terms of efficiency to destination and driving performance. In addition, drivers could save time and gasoline using a portable navigation system when in an unfamiliar region, and driving performance may be safer, despite the fact that the display screen of the phone is small. [Copyright &y& Elsevier]

• This paper utilizes novel connected vehicle (CV) emulated data to predict real-time crash potential for arterials. • The CV emulated data are flexible to obtain. Different speed-related variables are estimated based on the CV emulated data to depict continuous traffic conditions • The results proved the feasibility of using CV emulated data for real-time crash potential prediction • The proposed methods can be applied to other types of vehicles Real-time crash potential prediction could provide valuable information for Active Traffic Management Systems. Fixed infrastructure-based vehicle detection devices were widely used in the previous studies to obtain different types of data for crash potential prediction. However, it was difficult to obtain data in large range through these devices due to the costs of installation and maintenance. This paper introduced a novel connected vehicle (CV) emulated data for real-time crash potential prediction. Different from the fixed devices' data, CV emulated data have high flexibility and can be obtained continuously with relatively low cost. Crash and CV emulated data were collected from two urban arterials in Orlando, USA. Crash data were archived by the Signal for Analytics system (S4A), while the CV emulated data were obtained through the data collection API with a high frequency. Different data cleaning and preparation techniques were implemented, while various speed-related variables were generated from the CV emulated data. A Long Short-term Memory (LSTM) neural network was trained to predict the crash potential in the next 5−10 min. The results from the model illustrated the feasibility of using a novel CV emulated data to predict real-time crash potential. The average and 50th percentile speed were the two most important variables for the crash potential prediction. In addition, the proposed LSTM outperformed Bayesian logistics regression and XGBoost in terms of sensitivity, Area under Curve (AUC), and false alarm rate. With the rapid development of the connected vehicle systems, the results from this paper can be extended to other types of vehicles and data, which can significantly enhance traffic safety. [ABSTRACT FROM AUTHOR]

This editorial is both an introduction to the papers that make up this special issue (on the Relationship between Crash Causation, Countermeasures, and Policy) and an attempt at drawing conclusions. To assist the reader, we begin with a brief description of the subject matter of each paper. As expected, the authors tackle different aspects of this general topic and often differ in their conclusions. We follow up by asking: Are in-depth crash causation studies helpful? Can the need for understanding causation be defended? Does the Swiss Cheese Metaphor require revision? What are the building blocks on which the crash injury prevention programs rest? Can one really avoid comparing costs and benefits? These are some of the issues we raise and discuss. We end by offering for consideration a realistic model to link causes, countermeasures, policy, and responsibility for public safety. [ABSTRACT FROM AUTHOR]

• This paper proposes the Grid-Based Conflict Index (GBCI) to quantify the collision risks and severity among the vehicles. • This paper introduced a calculation method for the grid-based Post Encroachment Time (PET) and the total kinetic energy change. • This paper proposed a traffic-safety-based AIM model aimed at minimizing the weighted sum of total delay and conflict risk at the intersection. Existing studies on autonomous intersection management (AIM) primarily focus on traffic efficiency, often overlooking the overall intersection safety, where conflict separation is simplified and traffic conflicts are inadequately assessed. In this paper, we introduce a calculation method for the grid-based Post Encroachment Time (PET) and the total kinetic energy change before and after collisions. The improved grid-based PET metric provides a more accurate estimation of collision probability, and the total kinetic energy change serves as a precise measure of collision severity. Consequently, we establish the Grid-Based Conflict Index (GBCI) to systematically quantify collision risks between vehicles at an autonomous intersection. Then, we propose a traffic-safety-based AIM model aimed at minimizing the weighted sum of total delay and conflict risk at the intersection. This entails the optimization of entry time and trajectory for each vehicle within the intersection, achieving traffic control that prioritizes overall intersection safety. Our results demonstrate that GBCI effectively assesses conflict risks within the intersection, and the proposed AIM model significantly reduces conflict risks between vehicles and enhances traffic safety while ensuring intersection efficiency. [ABSTRACT FROM AUTHOR]

• How does driver behavior expressed through head pose dynamics relate to involvement in safety critical events. • Computer vision and mixed logit techniques used to identify patterns and relationships between drivers' head pose dynamics and crash involvement. • Angular deviation for head pose dynamics indicated by yaw, pitch and roll increase the likelihood of crash intensity by 4.56%, 4.92% and 8.26% respectively. This paper investigates the role of driver behavior especially head pose dynamics in safety–critical events (SCEs). Using a large dataset collected in a naturalistic driving study, this paper analyzes the head pose dynamics and driving behavior in moments leading up to crashes or near-crashes. The study uses advanced computer vision and mixed logit modeling techniques to identify patterns and relationships between drivers' head pose dynamics and crash involvement. The results suggest that driver-head pose dynamics, especially poses that indicate distraction and movement volatility, are important factors that can contribute to undesirable safety outcomes. Marginal effects show that angular deviation for head pose dynamics indicated by yaw, pitch and roll increase the likelihood of crash intensity by 4.56%, 4.92% and 8.26% respectively. Furthermore, traffic flow and lane changing also contribute to increase in likelihood of crash intensity. These findings provide new insights into pre-crash factors, especially human factors and safety–critical events. The study highlights the importance of considering human factors in designing driver assistance systems and developing safer vehicles. This research contributes by examining naturalistic driving data at the microscopic level with early detection of behaviors that lead to SCEs and provides a basis for future research on automation. [ABSTRACT FROM AUTHOR]

• To address the issues of inconsistency and loss of abrupt changes that existing methods fail to solve, we propose a bilevel optimization method. For a thorough resolution of the data inconsistency, we utilize the ℓ l 2 trend filter to merge the raw position and velocity data. To preserve abrupt changes in trajectory, we utilize the ℓ l 1 trend filter to make the trajectory physically feasible and preserve driving characteristics. • To validate the effectiveness of the proposed method, we design experiments and calculate the metrics to verify it. We discuss the trend filtering order and demonstrate that the optimization order we proposed is optimal through experiments. We also calculate metrics such as RMSE and jerk value and also compare the other trajectory enhancement methods and the result of the trajectory prediction model before and after our processing method. Our method achieves superior performance on most of them. This approach is essential to ensure the reliability and accuracy of the results. • We provide a detailed explanation of the algorithm workflow and present examples. In our work, we take the latest trajectory dataset: CitySim dataset as an example to show how our method processes data and its effectiveness in handling trajectory data. It can serve as a reference and enable researchers to easily understand the method. What's more, the processing methods proposed are also applicable to other trajectory datasets. Trajectory data play a vital role in the field of traffic research such as vehicle safety, traffic flow, and intelligent vehicles. The quality of trajectory data will determine the safety effectiveness of both research and practical applications. Effectively filtering out noise and errors from trajectory data is crucial for improving data quality and further research. However, most enhancement methods only focus on the smoothness of trajectory but overlook abrupt changes. The processed trajectory still exist issues such as incomplete elimination of inconsistency and loss of driving characteristics. In this paper, we propose a generic optimization-based enhancement method to address the issues above. We propose a bilevel optimization method combined with ℓ l 1 and ℓ l 2 trend filter. First, we design a l ℓ 2 trend filter to fuse raw trajectory data and eliminate the inconsistency. Next, we utilize the l ℓ 1 trend filter to optimize the data, ensuring physical feasibility and preserving abrupt changes (emergency driving characteristics). Then, we validate the effectiveness of the method through evaluation metrics and prediction models. The generic optimization-based enhancement method proposed in this paper ensures the safety of both research and application by providing high-quality trajectory data. [ABSTRACT FROM AUTHOR]

• This paper explores the performance of the NB-L model in hotspot identification. • A Monte Carlo simulation protocol is designed to generate various simulated scenarios. • The findings reveal a trade-off between the NB-L and NB models in hotspot identification. • For dispersed data, the NB-L exhibits better specificity, and the NB better sensitivity in hotspot identification. Identifying hazardous crash sites (or hotspots) is a crucial step in highway safety management. The Negative Binomial (NB) model is the most common model used in safety analyses and evaluations - including hotspot identification. The NB model, however, is not without limitations. In fact, this model does not perform well when data are highly dispersed, include excess zero observations, or have a long tail. Recently, the Negative Binomial-Lindley (NB-L) model has been proposed as an alternative to the NB. The NB-L model overcomes several limitations related to the NB, such as addressing the issue of excess zero observations in highly dispersed data. However, it is not clear how the NB-L model performs regarding the hotspot identification. In this paper, an innovative Monte Carlo simulation protocol was designed to generate a wide range of simulated data characterized by different means, dispersions, and percentage of zeros. Next, the NB-L model was written as a Full-Bayes hierarchical model and compared with the Full-Bayes NB model for hotspot identification using extensive simulation scenarios. Most previous studies focused on statistical fit, and showed that the NB-L model fits the data better than the NB. In this research, however, we investigated the performance of the NB-L model in identifying the hazardous sites. We showed that there is a trade-off between the NB-L and NB when it comes to hotspot identification. Multiple performance metrics were used for the assessment. Among those, the results show that the NB-L model provides a better specificity in identifying hotspots, while the NB model provides a better sensitivity, especially for highly dispersed data. In other words, while the NB model performs better in identifying hazardous sites, the NB-L model performs better, when budget is limited, by not selecting non-hazardous sites as hazardous. [ABSTRACT FROM AUTHOR]

• Use of multiple data sources can enable more complete and informative road injury statistics. • The papers reviewed lack consistency in what was reported and how. • Probabilistic methods are more common when linking administrative data on crash injuries. • Linkage studies can identify topics and case types that warrant more attention in road safety programs, commonly including vulnerable road users. This systematic review examines studies of traffic injury that involved linkage of police crash data and hospital data and were published from 1994 to 2023 worldwide in English. Inclusion and exclusion criteria were the basis for selecting papers from PubMed, Web of Science, and Scopus, and for identifying additional relevant papers using PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and supplementary snowballing (n = 60). The selected papers were reviewed in terms of research objectives, data items and sample size included, temporal and spatial coverage, linkage methods and software tools, as well as linkage rates and most significant findings. Many studies found that the number of clinically significant road injury cases was much higher according to hospital data than crash data. Under-estimation of cases in crash data differs by road user type, pedestrian cases commonly being highly under-counted. A limited number of the papers were from low- and middle-income countries. The papers reviewed lack consistency in what was reported and how, which limited comparability. [ABSTRACT FROM AUTHOR]

The extraction and analysis of driving style are essential for a comprehensive understanding of human driving behaviours. Most existing studies rely on subjective questionnaires and specific experiments, posing challenges in accurately capturing authentic characteristics of group drivers in naturalistic driving scenarios. As scenario-oriented naturalistic driving data collected by advanced sensors becomes increasingly available, the application of data-driven methods allows for a exhaustive analysis of driving styles across multiple drivers. Following a theoretical differentiation of driving ability, driving performance, and driving style with essential clarifications, this paper proposes a quantitative determination method grounded in large-scale naturalistic driving data. Initially, this paper defines and derives driving ability and driving performance through trajectory optimisation modelling considering various cost indicators. Subsequently, this paper proposes an objective driving style extraction method grounded in the Gaussian mixture model. In the experimental phase, this study employs the proposed framework to extract both driving abilities and performances from the Waymo motion dataset, subsequently determining driving styles. This determination is accomplished through the establishment of quantifiable statistical distributions designed to mirror data characteristics. Furthermore, the paper investigates the distinctions between driving styles in different scenarios, utilising the Jensen–Shannon divergence and the Wilcoxon rank-sum test. The empirical findings substantiate correlations between driving styles and specific scenarios, encompassing both congestion and non-congestion as well as intersection and non-intersection scenarios. • A quantitative method of driving performance through trajectory costs is proposed. • A general driving style extraction method for the large-scale dataset is proposed. • Driving styles of different traffic scenarios are significantly different. [ABSTRACT FROM AUTHOR]

• This review synthesizes the literature on the role of traffic conflicts in roundabout safety evaluation. • Current surrogate safety indicators are not appropriate for measuring roundabout conflicts involving in three or more road users. • The configuration and geometry elements of a roundabout are related to the types and frequency of the conflicts. • The conflicts involving more than two vehicles or vulnerable road users need more studies in future research. • Traffic signs, signals and technologies of intelligent connected vehicles should be used comprehensively to control roundabout conflicts. The roundabout is one type of at-grade intersection commonly seen in many countries. The evaluation of roundabout safety is usually counted on conflict analysis of the roundabout traffic due to random and limited records of real accidents. This paper surveyed published papers and reports that investigate the role of traffic conflicts in roundabout safety evaluation. It summarized the definitions and observation methods of roundabout conflicts and classified the attributing factors of roundabout conflicts and the countermeasures to control the conflicts. This study found that although unique traffic flow movements at roundabouts create special patterns of roundabout conflicts, the methods of roundabout conflict analysis used in most existing studies were inherited from the studies of highway or cross-intersection conflicts, including conflict definitions, conflict measurements, and thresholds of conflict severity. Special or improper designs of roundabout configurations or basic geometry elements could arouse roundabout conflicts. The most common vehicle-to-vehicle conflicts were entering-circulating conflicts, sideswipe conflicts, and exiting-circulating conflicts. The conflicts among vehicles and vulnerable road users (VRUs) easily evolved into serious collisions, but these conflicts did not get deserved attention in previous studies. Drivers' familiarity with roundabouts also affected road users' safety. Traffic signs and pavement markings were commonly used to control roundabout conflicts, while traffic signals were more effective methods for the roundabouts with uneven distribution of approaching traffic or high traffic volume. Based on the analysis of existing studies, this paper pointed out seven future directions of further research in term of conflict measurement, data collection, infrastructure and access management, geometry, drivers and VRUs, signal control, and vehicle control. [ABSTRACT FROM AUTHOR]

• Takeover activities caused increases in mental workload, skin conductance level, and heart rate of the drivers. • Secondary tasks could distract the drivers and make them less engaged in driving takeover activities. • Takeover scenarios affected the average physiological responses of the drivers during takeover periods. • The ranges of physiological responses during the takeover periods varied a lot across different individuals. By the year 2045, it is projected that Autonomous Vehicles (AVs) will make up half of the new vehicle market. Successful adoption of AVs can reduce drivers' stress and fatigue, curb traffic congestion, and improve safety, mobility, and economic efficiency. Due to the limited intelligence in relevant technologies, human-in-the-loop modalities are still necessary to ensure the safety of AVs at current or near future stages, because the vehicles may not be able to handle all emergencies. Therefore, it is important to know the takeover readiness of the drivers to ensure the takeover quality and avoid any potential accidents. To achieve this, a comprehensive understanding of the drivers' physiological states is crucial. However, there is a lack of systematic analysis of the correlation between different human physiological responses and takeover behaviors which could serve as important references for future studies to determine the types of data to use. This paper provides a comprehensive analysis of the effects of takeover behaviors on the common physiological indicators. A program for conditional automation was developed based on a game engine and applied to a driving simulator. The experiment incorporated three types of secondary tasks, three takeover events, and two traffic densities. Brain signals, Skin Conductance Level (SCL), and Heart Rate (HR) of the participants were collected while they were performing the driving simulations. The Frontal Asymmetry Index (FAI) (as an indicator of engagement) and Mental Workload (MWL) were calculated from the brain signals to indicate the mental states of the participants. The results revealed that the FAI of the drivers would slightly decrease after the takeover alerts were issued when they were doing secondary tasks prior to the takeover activities, and the higher difficulty of the secondary tasks could lead to lower overall FAI during the takeover periods. In contrast, The MWL and SCL increased during the takeover periods. The HR also increased rapidly at the beginning of the takeover period but dropped back to a normal level quickly. It was found that a fake takeover alert would lead to lower overall HR, slower increase, and lower peak of SCL during the takeover periods. Moreover, the higher traffic density scenarios were associated with higher MWL, and a more difficult secondary task would lead to higher MWL and HR during the takeover activities. A preliminary discussion of the correlation between the physiological data, takeover scenario, and vehicle data (that relevant to takeover readiness) was then conducted, revealing that although takeover event, SCL, and HR had slightly higher correlations with the maximum acceleration and reaction time, none of them dominated the takeover readiness. In addition, the analysis of the data across different participants was conducted, which emphasized the importance of considering standardization or normalization of the data when they were further used as input features for estimating takeover readiness. Overall, the results presented in this paper offer profound insights into the patterns of physiological data changes during takeover periods. These findings can be used as benchmarks for utilizing these variables as indicators of takeover preparedness and performance in future research endeavors. [ABSTRACT FROM AUTHOR]

• A road safety policy index consisting of 10 items is proposed. • The index is applied to evaluate road safety policy in Norway. • Different estimators of effect are compared. • Road safety policy has been effective in reducing fatal injury. The paper presents an exploratory study of a road safety policy index developed for Norway. The index consists of ten road safety measures for which data on their use from 1980 to 2021 are available. The ten measures were combined into an index which had an initial value of 50 in 1980 and increased to a value of 185 in 2021. To assess the application of the index in evaluating the effects of road safety policy, negative binomial regression models and multivariate time series models were developed for traffic fatalities, fatalities and serious injuries, and all injuries. The coefficient for the policy index was negative, indicating the road safety policy has contributed to reducing the number of fatalities and injuries. The size of this contribution can be estimated by means of at least three estimators that do not always produce identical values. There is little doubt about the sign of the relationship: a stronger road safety policy (as indicated by index values) is associated with a larger decline in fatalities and injuries. A precise quantification is, however, not possible. Different estimators of effect, all of which can be regarded as plausible, yield different results. [ABSTRACT FROM AUTHOR]

• Changes in the risk of road traffic injury in Norway from 1970 to 2022 are described. • Possible explanations of the changes are discussed. • The reporting of injuries in official statistics has declined over time. • A real decline in risk is found after adjusting for incomplete reporting. This paper describes changes in the risk of road traffic injury in Norway during the period from 1970 to 2022. During this period, the risk of fatal and personal injury declined by more than 70 % for most groups of road users. There are five main potential explanations of a decline in the risk of injury: (1) a reduced probability of accidents that have the potential for causing injury; (2) an improved protection against injury given that an accident has occurred; (3) improved medical care increasing the survival rate, given an injury (this would reduce the number of fatalities, but not the number of injuries); (4) a tendency for the reporting of injuries in official accident statistics to decline over time; (5) uncertain or erroneous estimates of the exposure to the risk of injury. The decline in the risk of road traffic injuries in Norway after 1970 can probably be attributed to a combination of reduced reporting of injuries in official statistics, improved protection against injury in accidents, and (for fatal injuries) improved medical care. Insurance data, available from 1992, do not indicate a reduction in the risk of accidents leading to insurance claims. Incomplete and possibly erroneous data for mopeds and motorcycles make it impossible to identify sources of changes in injury risk over time for these modes of transport. [ABSTRACT FROM AUTHOR]

• BHLS offers the lowest safety performance compared to BRT and arterial networks. • The greatest risk of fatalities, injuries, and property damage occurs in the BHLS. • BRT offers lower crash rates in less severe events as injuries and fatalities rise. • Factors such as signalized intersection density and curvature improve road safety. Public transport priority systems such as Bus Rapid Transit (BRT) and Buses with High Level of Service (BHLS) are top-rated solutions to mobility in low-income and middle-income cities. There is scientific agreement that the safety performance level of these systems depends on their functional, operational, and infrastructure characteristics. However, there needs to be more evidence on how the different characteristics of bus corridors might influence safety. This paper aims to shed some light on this area by structuring a multivariate negative binomial model comparing crash risk on arterial roads, BRT, and BHLS corridors in Bogotá, Colombia. The analyzed infrastructure includes 712.1 km of arterial roads with standard bus service, 194.1 km of BRT network, and 135.6 km of BHLS network. The study considered crashes from 2015 to 2018 –fatalities, injuries, and property damage only– and 30 operational and infrastructure variables grouped into six classes –exposure, road design, infrastructure, public means of transport, and land use. A multicriteria process was applied for model selection, including the structure and predictive power based on [i] Akaike information criteria, [ii] K-fold cross-validation, and [iii] model parsimony. Relevant findings suggest that in terms of observed and expected accident rates and their relationship with the magnitude of exposure –logarithm of average annual traffic volumes at the peak hour (LOG_AAPHT) and the percentage of motorcycles, cars, buses, and trucks– the greatest risk of fatalities, injuries, and property damage occurs in the BHLS network. BRT network provides lower crash rates in less severe collisions while increasing injuries and fatalities. When comparing the BHLS network and the standard design of arterial roads, BHLS infrastructure, despite increasing mobility benefits, provides the lowest safety performance among the three analyzed networks. Individual factors of the study could also contribute to designing safer roads related to signalized intersection density and curvature. These findings support the unique characteristics and traffic dynamics present in the context of Bogotá that could inform and guide decisions of corresponding authorities in other highly dense urban areas from developing countries. [ABSTRACT FROM AUTHOR]

• Research into hazard perception training has focused primarily on drivers. • Training was found to improve hazard perception with moderate to large effects. • Training should employ a method which actively engages the participants. • There is considerable heterogeneity in the training methods and measures used. • Future research is needed in long-term retention. Hazard perception (HP) has been argued to improve with experience, with numerous training programs having been developed in an attempt to fast track the development of this critical safety skill. To date, there has been little synthesis of these methods. Objective: The present study sought to synthesise the literature for all road users to capture the breadth of methodologies and intervention types, and quantify their efficacy. Data Sources: A systematic review of both peer reviewed and non-peer-reviewed literature was completed. A total of 57 papers were found to have met inclusion criteria. Results: Research into hazard perception has focused primarily on drivers (with 42 studies), with a limited number of studies focusing on vulnerable road users, including motorcyclists (3 studies), cyclists (7 studies) and pedestrians (5 studies). Training was found to have a large significant effect on improving hazard perception skills for drivers (g = 0.78) and cyclists (g = 0.97), a moderate effect for pedestrians (g = 0.64) and small effect for motorcyclists (g = 0.42). There was considerable heterogeneity in the findings, with the efficacy of training varying as a function of the hazard perception skill being measured, the type of training enacted (active, passive or combined) and the number of sessions of training (single or multiple). Active training and single sessions were found to yield more consistent significant improvements in hazard perception. Conclusions: This study found that HP training improved HP skill across all road user groups with generally moderate to large effects identified. HP training should employ a training method that actively engages the participants in the training task. Preliminary results suggest that a single session of training may be sufficient to improve HP skill however more research is needed into the delivery of these single sessions and long-term retention. Further research is also required to determine whether improvements in early-stage skills translate to improvements in responses on the road, and the long-term retention of the skills developed through training. [ABSTRACT FROM AUTHOR]

• Factors affecting injury severity in rollover crashes were explored. • Random parameters logit model with unobserved heterogeneity in means and variances. • Findings reveal temporal instability of model specifications across individual years. • Aggregate-to-component shift quantified using out-of-sample prediction. • Implications for mitigating the associated driver injury severity. Single-vehicle rollover crashes have been acknowledged as a predominant highway crash type resulting in serious casualties. To investigate the heterogeneous impact of factors determining different injury severity levels in single-vehicle rollover crashes, the random parameters logit model with unobserved heterogeneity in means and variances was employed in this paper. A five-year dataset on single-vehicle rollover crashes, gathered in California from January 1, 2013, to December 31, 2017, was utilized. Driver injury severities that were determined to be outcome variables include no injury, minor injury, and severe injury. Characteristics pertaining to the crash, driver, temporal, vehicle, roadway, and environment were acknowledged as potential determinants. The results showed that the gender indicator specified to minor injury was consistently identified as a significant random parameter in four years' models and the joint five-year model, excluding the 2016 crash model where the night indicator associated with no injury was observed to produce the random effect. Additionally, two series of likelihood ratio tests were conducted to assess the year-to-year and aggregate-to-component temporal stability of model estimation results. Marginal effects of explanatory variables were also calculated and compared to analyze the temporal stability and interpret the results. The findings revealed an overall temporal instability of model specifications across individual years, while there is no significant aggregate-to-component variation. Injury severities were observed to be stably affected by several variables, including improper turn indicator, under the influence of alcohol indicator, old driver indicator, seatbelt indicator, insurance indicator, and airbag indicator. Furthermore, the year-to-year and aggregate-to-component shift was quantified and characterized by calculating the differences in probabilities between within-sample observations and out-of-sample predictions. The overall results imply that continuing to expand and refine the model to incorporate more comprehensive datasets can result in more robust and stable injury severity prediction, thus benefiting in mitigating the associated driver injury severity. [ABSTRACT FROM AUTHOR]

• Developed a RAD generation framework to generate traffic crash characteristics. • Used a multi-layered decision process for developing a RAD dataset. • Can serve as a universal benchmark system for model frameworks in safety research. In this paper, a framework is outlined to generate realistic artificial data (RAD) as a tool for comparing different models developed for safety analysis. The primary focus of transportation safety analysis is on identifying and quantifying the influence of factors contributing to traffic crash occurrence and its consequences. The current framework of comparing model structures using only observed data has limitations. With observed data, it is not possible to know how well the models mimic the true relationship between the dependent and independent variables. Further, real datasets do not allow researchers to evaluate the model performance for different levels of complexity of the dataset. RAD offers an innovative framework to address these limitations. Hence, we propose a RAD generation framework embedded with heterogeneous causal structures that generates crash data by considering crash occurrence as a trip level event impacted by trip level factors, demographics, roadway and vehicle attributes. Within our RAD generator we employ three specific modules: (a) disaggregate trip information generation, (b) crash data generation and (c) crash data aggregation. For disaggregate trip information generation, we employ a daily activity-travel realization for an urban region generated from an established activity-based model for the Chicago region. We use this data of more than 2 million daily trips to generate a subset of trips with crash data. For trips with crashes crash location, crash type, driver/vehicle characteristics, and crash severity. The daily RAD generation process is repeated for generating crash records at yearly or multi-year resolution. The crash databases generated can be employed to compare frequency models, severity models, crash type and various other dimensions by facility type – possibly establishing a universal benchmarking system for alternative model frameworks in safety literature. [ABSTRACT FROM AUTHOR]

• In-vehicle telematics has been used to monitor driving behaviour and insurer risk. • Machine learning is the dominant method for analysis of in-vehicle telematics data. • Speeding, braking, and distance variables are the most useful telematics variables. • Currently the effects of telematics feedback on driving behaviour are unknown. • Future research should consider individual trip differences and driver differences. Road traffic deaths are increasing globally, and preventable driving behaviours are a significant cause of these deaths. In-vehicle telematics has been seen as technology that can improve driving behaviour. The technology has been adopted by many insurance companies to track the behaviours of their consumers. This systematic review presents a summary of the ways that in-vehicle telematics has been modelled and analysed. Electronic searches were conducted on Scopus and Web of Science. Studies were only included if they had a sample size of 10 or more participants, collected their data over at least multiple days, and were published during or after 2010. 45 relevant papers were included in the review. 27 of these articles received a rating of "good" in the quality assessment. We found a divide in the literature regarding the use of in-vehicle telematics. Some articles were interested in the utility of in-vehicle telematics for insurance purposes, while others were interested in determining the influence that in-vehicle telematics has on driving behaviour. Machine learning analyses were the most common forms of analysis seen throughout the review, being especially common in articles with insurance-based outcomes. Acceleration, braking, and speed were the most common variables identified in the review. We recommend that future studies provide the demographical information of their sample so that the influence of in-vehicle telematics on the driving behaviours of different groups can be understood. It is also recommended that future studies use multi-level models to account for the hierarchical structure of the telematics data. This hierarchical structure refers to the individual trips for each driver. [ABSTRACT FROM AUTHOR]

• Paper estimates a CMF for high friction surface treatment (HFST) applied to horizontal curves on undivided two-lane roads. • Propensity score matching is incorporated into EB before-after methodology to select more appropriate reference group. • Results indicate significant reductions in crash frequency (of all types considered) when applying HFST. Horizontal curves are locations that, as a result of the changing alignment, may be a contributing factor in roadway departure crashes. One low-cost countermeasure to mitigate crashes at these locations is the installation of the high friction surface treatment (HFST), which increases roadway friction and is intended to help keep drivers on the roadway when traversing a horizontal curve. This treatment has been implemented at numerous curves in Pennsylvania, but the overall safety effectiveness is not known. The purpose of this study is to estimate a suite of Crash Modification Factors (CMFs) for HFST applied to curve sections of undivided two-lane roadways. A novel combination of the empirical Bayes observational before-after study design and propensity score matching was used to estimate CMFs for multiple crash types, crash severities, and roadway settings (urban and rural). Propensity score matching was implemented to identify the most appropriate reference group to use within the empirical Bayes methodology. The results indicate that the installation of HFST is associated with a statistically significant decrease in all crash types and severities considered. [ABSTRACT FROM AUTHOR]

• The paper investigates the associations between real-time mobility and safety metrics. • Trajectory data collected from autonomous vehicles (AVs) is utilized in a multi-site Bayesian analysis. • The proposed safety assessment method uses Bayesian hierarchical spatial random parameter extreme value model (BHSRP). • The model can handle the limited availability and uneven distribution of conflict data and accounts for unobserved spatial heterogeneity. • Mobility LOS E represented the most hazardous condition for vehicle safety. • The generated insights can enable safety-mobility conscious road networks design and planning. The study presents a real-time safety and mobility assessment approach using data generated by autonomous vehicles (AVs). The proposed safety assessment method uses Bayesian hierarchical spatial random parameter extreme value model (BHSRP), which can handle the limited availability and uneven distribution of conflict data and accounts for unobserved spatial heterogeneity. The approach estimates two real-time safety metrics: the risk of crash (RC) and return level (RL), using Time-To-Collision (TTC) as conflict indicator. Additionally, a Risk Exposure (RE) index was developed to reflect the risk of an individual vehicle to travel through a corridor. In parallel, the mobility of corridor were assessed based on the highway Capacity manual methodology using real-time traffic data (Highway Capacity Manual, 2010). The study used a 440-hour AVs' dataset of a corridor in Palo Alto, California. After normalizing for each LOS representation in the dataset, LOS E was identified as the most hazardous operating condition with the highest average crash risk. However, the time spent under different operating condition would affect the safety of individual vehicles traveling through a road facility (i.e., vehicle's exposure time). Accounting for exposure time, the vehicle has the highest chance of encountering an extremely risky driving condition at intersections and segments under LOS D and E, respectively. [ABSTRACT FROM AUTHOR]

• A dynamic test scenario generation method for AVs is proposed in this paper. The proposed method has the ability to generate a realistic driving environment and apply to more complex scenarios like lane changing scenarios, which is of significant value for AV testing and evaluation.** • Instead of reconstructing expert behavior based on the assumption of single modality, this method combines Hierarchical Dirichlet Process Hidden Semi-Markov model (HDP-HSMM) and GAIL, obtains the main modes through clustering, and directs the scenario generation process by conditioning the model on scenario class labels, which improves the scenarios diversity and generation efficiency. • A typical lane-changing scenario is used for the evaluation of the proposed method. The results show that this method can generate rich test scenarios, and test AVs' ability to deal with different kinds of dynamic scenarios. Autonomous vehicles must be comprehensively evaluated before deployed in cities and highways. However, most existing evaluation approaches for autonomous vehicles are static and model environmental vehicles with predefined trajectories, which ignore the time-sequential interactions between the ego vehicle and environmental vehicles. In this paper, we propose a dynamic test scenario generation method to evaluate autonomous vehicles by modeling environmental vehicles as agents with human behavior and simulating the interaction process between the autonomous vehicle and environmental vehicles. Considering the multimodal features of traffic scenarios, we cluster the real-word traffic environments, and integrate the scenario class labels into the conditional generative adversarial imitation learning (CGAIL) model to generate different types of traffic scenarios. The proposed method is validated in a typical lane-change scenario that involves frequent interactions between ego vehicle and environmental vehicles. Results show that the proposed method further test autonomous vehicles' ability to cope with dynamic scenarios, and can be used to infer the weaknesses of the tested vehicles. [ABSTRACT FROM AUTHOR]

• We identify co-occurrence patterns of attributes related to crashes and interventions • Latent Dirichlet allocation was applied to analyse the topics of RSI reports • Results reveal associations among conditions that have previously gone unreported • Results aid to determine if road authorities can implement compliance with standards Under the Safe System framework, Road Authorities have a responsibility to deliver inherently safe roads and streets. Addressing this problem depends on knowledge of the road network safety conditions and the number of funds available for new road safety interventions. It also requires the prioritisation of the various interventions that may generate benefits, increasing safety, while ensuring that reasonable steps are taken to remedy the deficiencies detected within a reasonable timeframe. In this context, Road Safety Inspections (RSI) are a proactive tool for identifying safety issues, consisting of a regular, systematic, on-site inspection of existing roads, covering the whole road network, carried out by trained safety expert teams. This paper aims to describe how topic modelling can be effectively used to identify co-occurrence patterns of attributes related to the run-off-road crashes, as well as the corresponding patterns of road safety interventions, as described in the RSI reports. We apply latent Dirichlet allocation (LDA), a widespread method for fitting a topic model, to analyse the topics mentioned in RSI reports, divided into two groups: problems found; and proposed solutions. For this study, 54 RSI gathered over six years (2012–2017) were analysed, covering 4011 km of Irish roads. The results indicate that important keywords relating to the "forgiving roadside" and "clear zone" concepts, as well as the relevant European technical standards (CEN-EN1317 and EN 12,767), are absent from the extracted latent topics. We also found that the frequency of topics related to roadside safety is higher in the problems record set than in the solutions record set, meaning that problems are more easily identified and related to the roadside area than interventions may be. This paper presents methodological empirical evidence that the LDA is appropriate for identifying the co-occurrence patterns of attributes related to the ROR crashes in road safety inspections' reports, as well as the interventions' patterns associated with these crashes. Also, it provides valuable information aimed to determine the extent to which national road authorities in Europe and their contractors are currently capable of implementing and maintaining compliance with roadside standards and guidelines throughout the life cycle of roads. [ABSTRACT FROM AUTHOR]

It is well established that angry and, subsequently, aggressive drivers pose a problem for road safety. Over recent years, there has been an increase in the number of published studies examining driver anger, particularly using the Driving Anger Scale (DAS). The DAS measures six broad types of situations likely to provoke anger while driving (i.e., police presence, illegal driving, discourtesy, traffic obstructions, slower drivers, and hostile gestures). The majority of the recent studies have moved away from traditional paper-and-pencil methodologies, using the internet to collect data, for reasons of convenience. However, it is not yet completely clear whether data obtained from this methodology differs from more traditional methods. While research outside of the driving arena has not found substantial differences, it is important to establish whether this also applies to driving-related research and measures, such as the DAS. The present study used Multigroup Confirmatory Factor Analysis (MGCFA) to investigate the invariance of the DAS across a random sample from the electoral roll (n = 1,081: males = 45%) and an internet sourced sample (n = 627; males = 55%). The MGCFA showed the same six-factor solution was supported in both datasets. The relationships between the DAS factors and age, sex, trait anger, and annual mileage were broadly similar, although more significant differences were identified in the internet sample. This research demonstrates that driving measures administered over the internet produce similar results to those obtained using more traditional methods. [ABSTRACT FROM AUTHOR]

• Developed spatial models for analyzing taxi-involved crashes in urban areas. • Leveraged massive taxi trip data for measuring spatial relationship between zones. • Used taxi traveled distance as an exposure variable in crash frequency modeling. • Enhanced performance of crash models with informative spatial autocorrelation. Maintaining taxi safety is one of the important goals of operating urban transportation systems. Taxicabs are often prone to higher crash risk due to their long-time exposure to the complicated and dynamic traffic environments in urban areas. Despite existing efforts in understanding the safety issues associated with these vehicles, there were still few attempts that have specifically examined the relationship between taxi-involved crashes and other multifaceted contributing factors. To this end, this paper aims to develop crash frequency models for analyzing taxi-involved crashes. In particular, the spatial autocorrelations between variables were explored and the Poisson conditional autoregressive (Poisson-CAR) models for taxi-involved crashes were proposed. Unlike previous safety studies that mainly consider distance as the key indicator of spatial correlation, the present paper introduced the use of massive taxi trip data for constructing a more informative spatial weight matrix. The developed models with the taxi trip-based weight matrix were tested by using the 2016 taxi trip data collected in Washington D.C. The modeling results highlight the key explanatory factors such as road density, taxi activity, number of bus stops, and land use. More importantly, it demonstrates that the proposed Poisson-CAR models with the taxi trip-based weight matrix outperformed both the non-spatial Poisson model and the Poisson-CAR models using conventional distance-based weight matrix. Moran's I tests further indicate that our proposed models have sufficiently accounted for the spatial autocorrelation of the residuals. Thus, it deserves to consider informative spatial weight matrices when applying spatial models in traffic safety studies. [ABSTRACT FROM AUTHOR]

• The environmental contributing factors to elderly pedestrian collision differ from ones to younger pedestrian collision. • Raised medians, 3-way intersections, and street trees make a positive contribution to the safety of elderly pedestrians. • Bus stops increase the odds of an intersection to be a hotspot of elderly pedestrian collisions. • The current road system designed for young, healthy users should be refurbished for older, vulnerable pedestrians. For the elderly, walking is an important, reliable mobility option, since the elderly frequently lose their physical and/or sensory ability to drive as their age increases. However, elderly pedestrians are vulnerable on the streets and are at great risk of injury or death, when involved in a collision. This is due to not only increased frailty but also such issues as reaction speed and confidence on the streets. Therefore, pedestrian safety for older adults is a growing concern. This paper comprehensively examines the relationship between physical conditions and elderly pedestrian safety at the intersection level. By constructing a multinomial logistic regression (MLR) model, this paper identifies the exclusive contributing factors to elderly pedestrian collisions rather than younger pedestrian collisions. The outputs from the model suggest that facilities such as raised median, three-way intersection, street tree, and park and recreational land use improve the safety of elderly pedestrians. They also imply that bus stops increase elderly pedestrian collisions, while the intersections with crosswalks or colored crosswalks do not contribute to elderly pedestrians' safety, but the safety of younger pedestrians. The findings of this paper provide insight to transportation policies like Complete Street and Vision Zero and help to improve the current road system that are designed for automobiles and young, healthy road users. [ABSTRACT FROM AUTHOR]

• It is essential to use a common definition to compare serious road traffic injury rates. • For estimating the number of serious injuries, at some point it is always necessary to use hospital records. • This paper explores the implications for in/exclusion criteria applied to case selection and. • Provides a methodological approach for converting ICD to MAIS codes. • Weighting factors could be used to correct data deviations and make more real estimations. To determine accurately the number of serious injuries at EU level and to compare serious injury rates between different countries it is essential to use a common definition. In January 2013, the High Level Group on Road Safety established the definition of serious injuries as patients with an injury level of MAIS3+(Maximum Abbreviated Injury Scale). Whatever the method used for estimating the number or serious injuries, at some point it is always necessary to use hospital records. The aim of this paper is to understand the implications for (1) in/exclusion criteria applied to case selection and (2) a methodological approach for converting ICD (International Classification of Diseases/Injuries) to MAIS codes, when estimating the number of road traffic serious injuries from hospital data. A descriptive analysis with hospital data from Spain and the Netherlands was carried out to examine the effect of certain choices concerning in- and exclusion criteria based on codes of the ICD9-CM and ICD10. The main parameters explored were: deaths before and after 30 days, readmissions, and external injury causes. Additionally, an analysis was done to explore the impact of using different conversion tools to derive MAIS3 + using data from Austria, Belgium, France, Germany, Netherlands, and Spain. Recommendations are given regarding the in/exclusion criteria and when there is incomplete data to ascertain a road injury, weighting factors could be used to correct data deviations and make more real estimations. [ABSTRACT FROM AUTHOR]

• Crash data used in safety analyses often contain erroneous or missing information. • Crash location, time and severity are the most frequently misreported attributes. • The impact of misreporting on crash analyses may be significant but not yet known. • Intelligent crash reporting systems are proposed for higher data reliability. Crash databases are one of the primary data sources for road safety research. Therefore, their quality is fundamental for the accuracy of crash analyses and, consequently the design of effective countermeasures. Although crash data often suffer from correctness and completeness issues, these are rarely discussed or addressed in crash analyses. Crash reports aim to answer the five "W" questions (i.e. When?, Where?, What?, Who? and Why?) of each crash by including a range of attributes. This paper reviews current literature on the state of crash data quality for each of these questions separately. The most serious data quality issues appear to be: inaccuracies in crash location and time, difficulties in data linkage (e.g. with traffic data) due to inconsistencies in databases, severity misclassification, inaccuracies and incompleteness of involved users' demographics and inaccurate identification of crash contributory factors. It is shown that the extent and the severity of data quality issues are not equal between attributes and the level of impact in road safety analyses is not yet entirely known. This paper highlights areas that require further research and provides some suggestions for the development of intelligent crash reporting systems. [ABSTRACT FROM AUTHOR]

• A "traverse" is the passage over a level crossing by a road user. • New data on traverses in Britain are used to give fatality rates per traverse. • Fatality rates for pedestrians are ten times higher than for road vehicle users. • Fatality rates at passive crossings are 100 times higher than at controlled crossings. • The valuation of delays and of fatalities at active crossings are compared. This paper investigates relationships between traverses, delays and fatalities to road users at railway level crossings in Great Britain. A 'traverse' means a passage across a level crossing by a road user, who may be a pedestrian, cyclist, or occupant of a road vehicle. The paper finds that the road users with the highest fatality rate per traverse are pedestrians at passive crossings. Their rate is about three orders of magnitude higher than that of users with the lowest risk, who are road vehicle occupants at railway-controlled crossings. The paper considers the choice between automatic and railway-controlled crossings on public roads. Railway-controlled crossings are widely used in Britain. They are about one order of magnitude safer than automatic crossings, but they impose greater delays on users. A formula is developed to give the overall delay to road users at either type of crossing in terms of the numbers of road users and trains per day, and in terms of the length of time that the crossing must be closed to the road to allow the passage of one train. It is found that automatic level crossings cause substantially less delay than railway-controlled level crossings. The official monetary values of road user delay and of preventing a fatality were used to estimate the valuations of delays and fatalities at hypothetical but representative automatic and railway-controlled crossings. These valuations were then used to explore the effect of replacing representative railway-controlled with automatic crossings or vice-versa. It is found that the valuation of the reduced delays from adopting automatic crossings typically outweighs the valuation of the losses from the increased casualties. However, in practice Britain has chosen to retain a large number of railway-controlled crossings, which implies accepting the delays in return for a good level crossing safety record. Finally, an analysis is carried out to determine the additional risk of typical car and walk journeys that involve traversing a level crossing compared with similar journeys that do not. It is found that the additional risk is small for motor vehicle journeys, but substantial for walk journeys. [ABSTRACT FROM AUTHOR]

• The role of human factors in the occurrence of accidents is presented. • Engineering ethics as an element of human factors is proposed. • The Challenger and Columbia space accidents as two case studies are discussed. • Negligence of engineering ethics as a gap in accident analysis models is described. • Ethical control is suggested in system boundary in systemic models in 3 levels. The purpose of this paper is to further investigate engineering ethics and its gap within accident analysis models. In this paper, at first, the role of human factors in the occurrence of accidents is presented. Then engineering ethics as an element of human factors is proposed. It is suggested that engineering ethics can provide engineers with the necessary guidelines to avoid possible accidents arising from their decisions and actions. In addition, the Challenger and Columbia space shuttle case studies that demonstrate the role of engineering ethics in the prevention and occurrence of accidents are discussed. Then sequential, epidemiological, and systemic accident analysis models are briefly investigated and negligence of engineering ethics as a gap in the accident analysis models is described. At the end, we suggest that by implementing engineering ethics as a controller within the system boundary in systemic accident models we may be able to identify and prevent the ethical causes of accidents. [ABSTRACT FROM AUTHOR]

• AV disengagement on public roads dominated by causes due to a planning issue. • AV disengagement induced by lacking certain numbers of radar and LiDAR sensors. • Drivers' take-over time impacted by disengagement cause and roadway characteristics. • To improve AV disengagements five or more radar sensors are needed. • Optimal number of LiDAR sensors to be installed is 3 or 4. Automated vehicles (AV) testing on the public roads is ongoing in several states in the US as well as in Europe and Asia. As long as the automated vehicle technology has not achieved full automation (Level 5), human drivers are still expected to take over the steering wheel and throttles when there is an automated vehicle disengagement. However, contributing factors and the mechanism about automated vehicle-initiated disengagement has not been quantitatively and comprehensively explored and investigated due to the lack of field test data. Besides, understanding human drivers' perception and promptness of reaction to the AV disengagement is essential to ensure safety transition between automated and manual driving. By harnessing California's Autonomous Vehicle Disengagement Report Database, which includes the AV disengagement data from field tests in 2016–2017, this paper quantitatively investigated the AV disengagement using multiple statistical modeling approaches that involve statistical modeling and classification tree. Specifically, the paper identifies the contributing factors impacting human drivers' promptness to AV disengagements, and quantitatively investigates the underlying causes to AV disengagements. Results indicate that current AV disengagement on public roads is dominated by causes due to a planning issue. The cause of an AV disengagement is significantly induced by lacking certain numbers of radar and LiDAR sensors installed on the automated vehicles. These thresholds of these sensors needed are revealed. Cause of disengagement and roadway characteristics significantly impact drivers' take-over time when facing an AV disengagement. AV perception or control issue-based disengagement can significantly extend drivers' perception-reaction time to take over the driving. The quantitative knowledge obtained ultimately facilitates revealing the mechanisms of the automated vehicle disengagements to ensure safe AV operations on public roads. [ABSTRACT FROM AUTHOR]

• Here are the highlights to this paper: • Classify various cyberattaks injection modes. • Propose three car-following models to analyze the cyberattack effects on platoon. • Apply six indicators to comprehensively evaluate driver tolerance, vehicle adaptability, and environmental resistance. The high-level integration and interaction between the information flow at the cyber layer and the physical subjects at the vehicular layer enables the connected automated vehicles (CAVs) to achieve rapid, cooperative and shared travel. However, the cyber layer is challenged by malicious attacks and the shortage of communication resources, which makes the vehicular layer suffer from system nonlinearity, disturbance randomness and behavior uncertainty, thus interfering with the stable operation of the platoon. So far, scholars usually adopt the method of assuming or improving the car-following model to explore the platoon behavior and the defense mechanism in cyberattacks, but they have not considered whether the model itself has disturbance and impact on cyberattack defenses. In other words, it is still being determined whether the car-following model designed can be fully applicable to such cyberattacks. To provide a theoretical basis for vehicular layer modeling, it is necessary to comprehend the self-resistance of different car-following models faced on various cyberattacks. First, we review the car-following models adopted on the vehicular layer in cyberattacks, involving traffic engineering, physical statistics, and platoon dynamics. Based on the review, we divide the malicious attacks faced by the cyber layer into explicit attacks and implicit attacks. Second, we develop a cooperative generalized force model (CGFM), which combines and unifies the r-predecessors following communication topology. The proposed models, labeled the vulnerable cooperative intelligent driver model (VCIDM), the vulnerable cooperative optimal velocity model (VCOVM), and the vulnerable cooperative platoon dynamics model (VCPDM), incorporate the CGFM model and assorted cyberattack injection modes to explain the cyberattack effects on the platoon self-resistance capability. Upon the described models, we provide six indicators in three dimensions from the basic traffic element, including drivers, vehicles, and environment. These indicators illustrate driver tolerance, vehicle adaptability, and environmental resistance when a platoon faces attacks such as bogus information, replay/delay, and communication interruption. We arrange and reorganize the car-following models and the cyberattack injection modes to complete the research on the self-resistance capability of the platoon, which has positive research value and practical significance for enhancing the endogenous security at the vehicular layer and improving the intrusion tolerability at the cyber layer. [ABSTRACT FROM AUTHOR]

• Commercial vehicles were the most likely to make a full stop and stop at the stop bar. • Correlations between roadway infrastructure (e.g. lighting) illustrate countermeasures with more positive safety impacts. • Drivers were much more likely to stop at the stop sign/bar compared to junctions without any additional safety measures. • Driver behavior data helps build a robust road safety database, improve policy, and create innovative informed initiatives. • Stopping behavior likelihoods can avail in selecting tailored solutions and predict the economical effectiveness of them. Ghana exemplifies the contribution of road crashes to mortality and morbidity in Africa, partly due to a growing population and increasing car ownership, where fatalities have increased by 12 to 15 % annually since 2008 (National Road Safety Authority (NRSA), 2017). The study described in this paper focused on understanding driver behavior at unsignalized junctions in the Ashanti Region of Ghana. Understanding driver behavior at unsignalized junctions is particularly important since failure to stop or yield can seriously affect vulnerable road users. The study's objectives were to develop relationships between driver behavior and junction characteristics. Understanding the characteristics that lead to determining what factors influence a driver's behavioral response at rural junctions provides information for policy makers to determine the best strategies to address these behaviors. The study evaluated stopping behavior at rural junctions. Driver behavior was extracted from video views of ten junctions in the Ashanti Region of Ghana. A total of 3,420 vehicles were observed across all ten junctions during data collection before any analysis was conducted. The type of stop was selected as a surrogate measure of safety. Logistic regression was used to model stopping behavior at the selected junctions. The analysis showed drivers were more likely to stop when going straight (versus a left turn) and left turning vehicles were more likely to stop than right turning vehicles. Additionally, single unit trucks and tro-tros were more likely to stop than other vehicle types. Drivers were also much more likely to stop when channelization, intersection lighting, or speed humps were present. Drivers at junctions with 4-approaches were also more likely to stop than those with 3 approaches. The results from this research contribute valuable information about what factors contribute to positive safety behaviors at rural junctions. This provides guidance for safety professionals to select solutions and can be a valuable tool to predict the economical effectiveness of solutions to addressing junction safety in low- and middle-income countries (LMIC) such as Ghana. The results can also provide insight and recommendations to Ghanaian road safety agencies and launch sustainable efforts to raise community awareness toward decreasing road crash fatalities in Ghana. [ABSTRACT FROM AUTHOR]

• A duration modeling approach of accumulated distances between crashes is used to evaluate automated vehicle safety. • Conventional vehicle crashes sourced from the naturalistic driving study are considered as the benchmark. • Statistical tests are conducted to statistically assess differences between automated and conventional vehicles. • The existing level 3 of automation is shown to be safer than conventional vehicles with 95 % confidence. • Automated vehicles have roughly 27 % more miles bewteen crashes than conventional. Ideally, the evaluation of automated vehicles would involve the careful tracking of individual vehicles and recording of observed crash events. Unfortunately, due to the low frequency of crash events, such data would require many years to acquire, and potentially place the motorized public at risk if defective automated technologies were present. To acquire information on the safety effectiveness of automated vehicles more quickly, this paper uses the collective crash histories of a group of automated vehicles, and applies a duration modeling approach to the accumulated distances between crashes. To demonstrate the applicability of this approach as a method compare automated and conventional vehicles (human drivers), an empirical assessment was undertaken using two comparable sources of data. For conventional vehicles, police and non-police-reportable crashes were collected from the Second Strategic Highway Research Program's naturalistic driving study, and for automated vehicles, data from the California Department of Motor Vehicles Autonomous Vehicle Tester program were used (105 crashes from 59 permit holders driving ∼2.8 million miles were used for the analysis). The results of the empirical study showed that automated driving was safer at the 95% confidence level, with a higher number of miles between crashes, relative to their conventional vehicle counterparts. The findings indicate that the number of miles between crashes would be increased by roughly 27% when switching from conventional vehicles to automated vehicles. Despite limited data which mandated a group-vehicle approach, this study can be considered a reasonable initial approximation of automated vehicle safety. [ABSTRACT FROM AUTHOR]

• Identifying the latent relationships between factors associated with traffic crashes through Markov random fields: Graphical models disclosed relationship topologies of factors leading to severe crashes. • The crash mechanisms illustrated by using graphical representation of relationships. • High dimensionality arising in case of rare crash types can be handled by GLE. • Essential factors jointly acting towards crash occurrence can be identified, similar to a pathological examination. • Proposed approaches can assist in devising accurate and reliable prevention measures. Traffic safety field has been oriented toward finding the relationships between crash outcomes and predictor variables to understand crash phenomena and/or predict future crashes. In the literature, the main framework established for this purpose is based on constructing a modelling equation in which crash outcome (e.g., frequencies) is examined in relation to explanatory variables chosen based on the problem at hand. Despite the importance and success of this approach, there are two issues that are generally not discussed: 1) the latent relationships between factors associated with crashes are oftentimes not the focus of analysis or not observed; and 2) there are not many tools to make informed decisions on which variables might have an impact on the crash outcome and should be included in a safety model, particularly when observations are limited. To address these issues, this paper proposes the use of graphical models, namely a Markov random field (MRF) modelling, Bayesian network modelling, and a graphical XGBoost approach, to disclose relationship topologies of explanatory variables leading to fatal and incapacitating injury pedestrian crashes. The application of graph learning models in traffic safety has a high potential because they are not only useful to understand the mechanism behind the crash occurrence but also can assist in devising accurate and reliable prevention measures by identifying the true variable structure and essential factors jointly acting towards crash occurrence, similar to a pathological examination. [ABSTRACT FROM AUTHOR]

• Single-Vehicle and Multi-Vehicle Truck-Involved Crashes Injury Severities are investigated. • Transferability and temporal instability are assessed through comparison between within- and out-of-sample prediction. • Temporally unconstrained parameters and partially temporally constrained parameters model are compared. • The presence of temporal instability across both single and multi-truck collision risk determinants carries critical countermeasure translation opportunities. • Factors maintaining stability amidst change also carry long-term safety implications. Truck-involved crashes persist as a significant concern, yielding noteworthy human casualties and causing economic ramifications, particularly in developing countries. This paper aims to undertake a comprehensive analysis of the associated factors influencing injury severity in truck-involved crashes, with a particular emphasis on discerning variations between single-vehicle and multi-vehicle incidents, as well as accounting for heterogeneity and temporal stability. The data analysis involves a meticulous examination of crash data spanning the entirety of Thailand from 2017 to 2020. Employing three distinct levels of injury severities, namely PDO injury, moderate injury, and severe injury, the study employs a series of mixed logit models that account for unobserved heterogeneity in both means and variances. Results revealed significant instability in injury risk determinants over time among both single and multi-vehicle events. Aligning predictive assessments further spotlighted fluctuations in projected burdens across models and years – collectively underscoring the imperative to integrate temporal considerations into modeling and prevention. Several crash-type distinctions and priorities emerged. For single-truck events, key risks included roadway alignments and geometry, speeding, fatigue, and lighting conditions. However multi-truck collisions concentrated around exposure factors like highway traits, sightline limitations, and vulnerable road users. Ultimately, the technique permitted responsive countermeasure targeting and recalibration opportunities keyed to each crash form's evolving landscapes. While it is indeed noteworthy that several variables have exhibited instability in their effects, it is equally important to acknowledge the existence of certain variables that maintain a relative degree of temporal stability. This underscores their pivotal role in shaping the foundation of enduring strategies aimed at enhancing traffic safety in the long run. The multifaceted investigation constitutes an invaluable reference for diverse transportation stakeholders seeking to curb rising truck fatalities through evidence-based improvements in policy, engineering, usage protocols, and technologies. It provides a blueprint for nimble safety planning within complex modernizing road systems. [ABSTRACT FROM AUTHOR]

• Vision Zero is a new approach to traffic safety, initiated in Sweden. • Both the UN, WHO and EU have put traffic safety on the political agenda. • The Vision Zero approach stresses a systems approach where system designers have responsibilities alongside the road users. • About two-thirds of the fatal crashes in Sweden are attributed to everyday road users making everyday mistakes. • Less than 20% of the fatal crashes injuries in Sweden are related to deliberate and severe violations of traffic rules. • A combination of divided roads, sober driver and restrained occupants not speeding is estimated to generate a reduction of fatal crashes of 95%. • The search for an isolated cause of fatalities a result of road traffic crashes has no longer any substantial role in injury prevention. • Modern road safety preventative methods are based on stopping or mitigating a sequence of events in the most effective way. • These modern methods are, to a high degree, disconnected from the more traditional finding of singular crash causes. • It can be seen as counterproductive that the judicial system still concentrates on finding a single cause related to an individual road user. The search for common and serious single causes of road crashes naturally leads to a concentration on the road user. This is supported by a legal framework in the search for the main cause and the suspect for this cause. In prevention, we have for decades been more inclined to look for systematic improvements of all elements of the road transport system, and we direct the recommendations for actions towards system designers, organizations, products and services. In this paper the discussion about causation and prevention is broadened in the light of Vision Zero and its approach to prevention of serious and fatal injuries. We also discuss the Swedish judicial system and why the prevention approach has not been legislated or even generally accepted. Occupational health and safety legislation and road rules are compared, as well as how sustainability practices and reporting are tools to apply prevention where organizations have a natural sphere of influence that could mitigate deaths and serious injuries within value chains. It is recommended that we stop using the term causation as it is only directing actions in one direction. There is a risk that the focus on causation, in particular single causes, will deviate actions away from robust prevention countermeasures such as increased seat belt use, relevant speed limits, and well functioning roundabouts and median barriers. Furthermore, there is also a risk that important preventative actions from organizations are overlooked. [ABSTRACT FROM AUTHOR]

• More efficient traffic light preemption strategies and faster medical emergency response can be achieved by having a global view of the city traffic. • A traffic light preemption strategy is proposed taking advantage of software defined everything (SDX) concept, namely SD-TLP. • SD-TLP takes advantage of a centralized controller and thus improves the preemption performance. • SD-TLP achieves a rescue time that is close to optimal while ensuring that its impact on regular city traffic remains reasonably low. • Lane-changing mechanisms offer benefits in areas with light city traffic and can serve as a supplementary strategy to SD-TLP. Proper management of rescue operations following an accident is one of the most fundamental challenges faced by today's smart cities. Taking advantage of vehicular communications, in this paper we propose novel mechanisms for the acceleration of the rescue operation resulting in a reduction in fatalities in accidents. We propose a Software-Defined Traffic Light Preemption (SD-TLP) mechanism that enables Emergency Medical Vehicles (EMVs) to travel along the rescue route with minimal interruptions. The SD-TLP makes preemption decisions based on global knowledge of the traffic conditions in the city. We also propose mechanisms for the selection of the nearest emergency center and fast discharge of the route of EMVs. Furthermore, depending on the dynamic traffic conditions on the streets at the time of the accident, an appropriate rescue route is selected for the EMV before its departure. The proposed approach is evaluated using the OMNET++ and SUMO tools over part of the Megacity of Tabriz, Iran. The simulation results demonstrate that the method can reduce the average rescue time significantly. The proposed approach keeps the resulting disruption in city traffic acceptably low while trying to shorten the rescue time as much as possible. [ABSTRACT FROM AUTHOR]

• A bibliometric analysis of the state of art in Telematics-Based Automobile Insurance research. • Identification of annual scientific production, country-wise contribution and outlet performance. • An examination of the intellectual structure using cluster analysis. • Accident Analysis and Prevention was a notable journal on this topic. • Consumers' adoption behaviour and their privacy concerns must be investigated. Telematics technology and its implementation in auto insurance have received great interest due to their potential to transform the insurance sector and promote safer driving practices. By implementing telematics technology, insurers may tailor insurance premiums to individual drivers, taking into account their real driving habits and performance, ultimately leading to improved road safety, cost savings, and an empowered driving community. The current study, through bibliometric analysis, carefully identifies and evaluates the existing body of scholarly literature on this subject for the last 21 years, including journal articles, conference papers, and related publications. The analysis uncovers key research studies, influential authors, top publication outlets, top countries with collaborations, and prolific research fields, providing useful insights into the evolution and growth of telematics-based insurance research. Furthermore, thematic mapping, cluster analysis, and critical analysis of top recent studies aided in identifying key research clusters and themes, as well as potential gaps and areas for further exploration, guiding future researchers and policymakers in advancing this transformative technology. [ABSTRACT FROM AUTHOR]

• There are no comprehensive pre-post evaluations of bicycle passing distance rules. • The Queensland passing distance rule had high levels of awareness and compliance. • Countermeasure implementation sometimes precedes understanding of causation. • Countermeasure evaluations can generate data useful to understand crash causation. Close passes by motor vehicles endanger both the safety and comfort of bicycle riders. Governments in many countries have introduced laws requiring drivers to maintain at least a minimum distance between their vehicle and the cyclist they are passing, despite relatively poor understanding of the causes of bicycle overtaking crashes at the time. Queensland was the first state in Australia to introduce such a law, with a two-year trial commencing in April 2014. The data collected during the evaluation of the trial were later analysed to answer two main questions: "Under what circumstances do close passes occur?" and "Why do drivers pass too close?". The first question was largely approached by analysing the video observations of more than 18,000 riders (including 2,000 passing events) at 15 locations on Queensland roads and examining the infrastructure, traffic and road user characteristics that influenced passing distances. The second question was addressed in experimental studies which used the video observations as stimuli. This paper demonstrates how the political need for evaluation of a countermeasure can act as a stimulus for research funding that then allows data collection, analysis and better understanding of crash causation. Logically, introduction of a countermeasure should be based on a rigorous understanding of crash causation. But when this does not occur, evaluation may provide data that can be used to answer questions about crash causation – or at least pose new questions. [ABSTRACT FROM AUTHOR]

• Logistics-DT-TFP is proposed for analyzing the causes of RDGT accidents. • Ordered multi-classification Logistic regression is used to analyze cause and accident severity. • SMOTE-Tomek mixed sampling is used for data balance. • DT and C5.0 algorithm are used to extract and analyze the coupling rules. • TOP-K Improved FP-Growth is used to mine association and supplement coupling rules. Accurately and quickly mining the hidden information in railway dangerous goods transportation (RDGT) accident reports has great significance for its safety management. In this paper, a data mining method Logistics-DT-TFP is proposed for analysing the causes of RDGT accidents. Firstly, analyse the transportation process, extract the cause of the accident, and classify the severity of the accident. Then, using ordered multi-classification Logistic regression for correlation calculation, qualitatively judge and quantitatively analyse the relationship between each cause and the severity of the accident. The feature tags of the Decision Tree (DT) are screened, the C5.0 algorithm is used to obtain the accident coupling rules. Next, the FP-Growth algorithm is used to mine frequent itemsets, and TOP-K is used to improve it and output effective association rules with the degree of lift as the indicator, which avoids repeated traversal of the database, shortens the time complexity, and reduces the impact of the minimum support setting on the calculation results. The degree of lift among the causes in the coupling chain is calculated as a complement to the extraction of coupling rules. Finally, based on the analysis and mining results of case study, the management strategies for railway dangerous goods are proposed. [ABSTRACT FROM AUTHOR]

• Determinants of motorcyclists' injury severities are studied. • Methodological issues of scale and random heterogeneity are captured. • Mixed logit models considering generalized scale heterogeneity are estimated. • Random parameters logit model considering heterogeneity in means and vaeiances are also estimated. • Temporal instability warrants for estimation of separate year-specific models. • Significant differences in sources of heterogeneity over the years are observed. The presence of unobserved factors in the motorcycle involved two-vehicle crashes (MV) data could lead to heterogenous associations between observed factors and injury severity sustained by motorcyclists. Capturing such heterogeneities necessitates distinct methodological approaches, of which random and scale heterogeneity models are paramount. Herein, we undertake an empirical evaluation of random and scale heterogeneity models, exploring their efficacy in delineating the influence of external determinants on the degree of injury severity in crashes. Within the effects of scale heterogeneity, this study delves into two dominant models: the scaled multinomial logit model (S-MNL) and its generalized counterpart, the G-MNL, which encompasses both the S-MNL and the random parameters multinomial logit model (RPL). While the random heterogeneity domain is represented by the random parameters multinomial logit and an upgraded variant – the random parameters multinomial logit model with heterogeneity in means and variances (RPLHMV). Motorcycle involved two-vehicle crashes data were extracted from the UK STATS19 dataset from 2016 to 2020. Likelihood ratio tests are computed to assess the temporal variability of the significant factors. The test result demonstrates the temporal variations over a five-year study period. Some very important differences started to show up across the years based on the model estimation results: that the RPLHMV model statistically outperforms the G-MNL model in the 2016, 2018, and 2019 models, while the S-MNL model is statistically superior in the 2017 and 2020 years. These important findings suggest that the origin of heterogeneity in explaining factor weights can be captured by scale effects, not just random heterogeneity. In addition, the model results further show that motorcyclists' injury severities are significantly affected by motorcycle-related characteristics; there is the added factor of external influences, such as non-motorcycle drivers (males, young drivers, and elderly drivers) and vehicles (the moving status, age, and types of vehicles) that collide with motorcycles. The results of this paper are anticipated to help policymakers develop effective strategies to mitigate motorcycle involved two-vehicle crashes by implementing appropriate measures. [ABSTRACT FROM AUTHOR]

• Accident modification functions for traffic law enforcement are compared. • All functions show that more enforcement improves safety. • Differences-in-differences estimates capture both increases and decreases in enforcement. • A decrease in enforcement worsens road safety. Traffic law enforcement is a road safety measure whose effects on accidents or injuries is best described by means of a function rather than a point estimate. An informative function should comprise both increases and decreases in enforcement. Currently available accident modification functions cannot serve this need. A fruitful approach to developing accident modification functions covering both increases and decreases in enforcement is differences-in-differences estimates based on multivariate accident prediction models. The paper explains how to develop such estimates and illustrates them. The interpretation of the results of empirical studies can be informed by a game-theoretic model of the effects of enforcement, previously published in Accident Analysis and Prevention (Bjørnskau and Elvik 1992, 507–520). [ABSTRACT FROM AUTHOR]

• A novel visualisation for a road safety evidence-based policy making framework is presented. • 3 pillars considered: Crash causation; Countermeasures/implementation; Road safety management. • A summary of previous research relating to the 3 pillars is provided. • Framework is a high-level map of best practice for countries to improve their evidence-based policy. Effective evidence-based policy making within road safety is a several step cyclic process that involves gathering data about the causes of crashes, analysing these data, developing countermeasures and implementing and evaluating them. There are many examples of crash causation focused data collection activities available to policy makers but knowledge on how these finding may have led to countermeasure implementation and new policy is much less well established. This paper proposes a framework for best practice evidence-based policy making. To address existing gaps, the framework consists of three pillars: these are (1) Crash causation establishment; (2) Countermeasure development and implementation; and (3) road safety management. A key element in this framework is the recommendation for the establishment of an organisation responsible for road safety that has a strategic and coordination role. This framework, as a whole, aims to provide a practical high-level map by connecting evidence to policy at every point in the policy making cycle and ensuring that evidence-based road safety policy is a national priority. It is anticipated that using this framework to inform road safety policy development will enhance the success of any developed policy. [ABSTRACT FROM AUTHOR]

Road safety has become a global concern but its impact in low- and middle-income countries is widespread mainly due to lack of appropriate crash database system and under-reporting. In this context, the primary objective of this paper is to provide a scalable framework for unveiling pedestrians' perceived road safety that can also be applied in regions where accessible crash data are limited or near-crashes are left unreported. In the first step of our methodology, a deep learning architecture-based semantic segmentation model (HRNet+OCR) is trained using labeled Google Street View (GSV) images from specific study areas in Dhaka, Bangladesh, which facilitates the identification of both man-made components (such as roads, sidewalks, buildings, and vehicles) and natural elements (including trees and sky). The developed model showed excellent performance in identifying different features in an image by achieving high precision (0.95), recall (0.97), F1-score (0.96), and intersection over union (IoU) (91.86). Secondly, a group of trained raters scored the perceived road safety on an ordinal scale from 0 to 10 (extremely unsafe to extremely safe to walk in terms of road crashes) by assessing the GSV images. Then, several regression models have been used on features extracted from GSV images, and socio-demographic factors (i.e., population density, and relative wealth index) to estimate the perceived road safety, and random forest regression model was found to perform the best. Further, Shapley Additive Explanations (SHAP), a model-agnostic technique has been used for examining feature importance by computing the contribution of each feature to the random forest regression model output. The results show that sidewalk, road, population density, wall, and relative wealth index have higher impact on determining the perceived road safety rating. Additionally, the results of t-tests between the average perceived road safety scores for crash-prone and non crash-prone areas revealed the existence of significant differences. This study also provides perceived road safety rating map on a neighborhood scale, which can be a useful visualization tool for policy-makers and practitioners to identify the road safety deficiencies at specific locations, and formulate appropriate and strategic countermeasures to improve pedestrians' road safety. • A scalable framework for assessing pedestrian's road safety perception. • Unveiling natural and built-environment factors affecting road safety perception. • Semantic segmentation model performed well to identify man-made and natural features. • Differences exist between perceived road safety and reported crash data. • Perceived road safety rating map will aid traffic safety stakeholders. [ABSTRACT FROM AUTHOR]

• E-scooter riders are more likely to engage in illegal riding behaviour than non-riders. • E-scooter riders and non-riders have limited knowledge of e-scooter legislation. • Awareness of legislation correlates positively with law-abiding e-scooter practices. • Training and educational campaigns are needed to improve riders' safety behaviour. The use of e-scooters is rapidly increasing in cities, leading to their integration into the transportation system. However, numerous collisions involving e-scooters, including some resulting in fatalities, have been reported since their introduction. These incidents indicate that the potential dangers posed by e-scooters may be underestimated. Research suggests that a significant factor contributing to these collisions is the prevalence of illegal riding behaviour exhibited by many riders. This paper presents three studies that aimed to assess the understanding of e-scooter riders and non-riders of the current legislation across various riding scenarios and link it to their profile, riding habits, and their proneness to engage in illegal riding behaviours. Study 1 utilised questionnaires to survey participants and gather information about their profiles and self-reported illegal riding behaviour. Study 2 focused on assessing participants' knowledge of the current e-scooter legislation through different everyday riding scenarios. Study 3 featured short video clips from the rider's perspective to determine the proneness of participants to engage in illegal riding behaviour and explore the potential relationship between these behaviours and their understanding of e-scooter rules. The findings revealed that e-scooter riders were generally younger and exhibited a higher propensity for engaging in illegal riding behaviour than non-users. Both groups demonstrated limited knowledge regarding various aspects of the current e-scooter legislation, particularly related to parking, speeding, and designated infrastructure. While e-scooter riders demonstrated relatively greater knowledge of the e-scooter rules, this was not consistently observed across all areas, particularly in relation to riding on pavements (pedestrian footpaths). Furthermore, Study 3 revealed that participants with better knowledge of the current legislation were less likely to engage in illegal riding behaviour. These findings suggest a need for targeted interventions and educational campaigns to improve riders' understanding of regulations and promote safer riding practices. Implementing training programs for e-scooter safety could significantly enhance riders' awareness of the associated dangers, fostering responsible e-scooter use. [ABSTRACT FROM AUTHOR]

• 'Best practices' to analyse bicycle crash causes in correlational studies. • An abbreviated list of crash causation criteria was used. • Risk comparisons between two types of bicycles is the example given. • Controlling for age, gender, and exposure is required to analyse crash risk. The increasing number of bicycle crashes leading to injuries in the Netherlands is frequently related (e.g., in the media) to increased use of the electric bicycle. For this reason, policy makers face the challenge of selecting and implementing the most promising countermeasures including those focused on electric bicycles. It may be questioned, however, to what extent the electric bicycle itself is a (direct) cause of crashes leading to injuries or whether other factors are important for explaining the increased number of bicycle injury crashes. On the basis of an abbreviated list of criteria by Elvik (2011) , this paper illustrates the potential influence of factor inclusions, analysis selections, and data presentations on the general impression about crash causation with the electric bicycle as an example. The aim is to provide a 'best practice guide' by taking into account (1) a theoretical explanation of causal mechanisms, (2) control for confounders, and (3) a statistical association of sufficient strength and consistency in the expected direction. We conclude that an apparent increased risk of electric bicycles may be explained by factors such as age, exposure, health factors, and gender of the cyclist. A relatively simple analysis, by comparing fatality numbers of conventional and electric bicycles, showed that including or excluding these factors may lead to vastly different interpretations of fatality causes and the relative risk of electric bicycles compared to conventional bicycles. [ABSTRACT FROM AUTHOR]

• Long short-term memory can be applied to predict risks related to driving on sharp curves. • The effects of road alignment and driving characteristics on driving risks can be considered using the dynamic and spatial classification of curves. • Model prediction performance is affected by observation and delay windows. • Model performance can be improved by combining the data for vehicle kinematics, driver maneuvering/behavior, driver physiological characteristics, and road alignment characteristics. Traffic accidents are likely to occur on sharp curves under poor driving conditions, and the severity level of such accidents is high. Therefore, predicting the risk associated with driving on curved roadways in real time can effectively improve driving safety. This paper aims to develop a dynamic real-time method that fuses multiple data sources to predict risk when driving on sharp curves in the context of the connected vehicle environment. Six curves with three small radii (60 m, 100 m, 150 m) and two driving directions (left and right) were designed for a driving simulation experiment. Driver maneuvering data, vehicle kinematic data, and physiological data of 55 drivers were collected for this study. The data were combined and spatially and dynamically segmented. The mean value of the critical lateral acceleration of the vehicle was set as the risk assessment index. K-means clustering was used to classify the driving risk into three levels: low, medium, and high. Then, the risk level was predicted using the maneuvering data, vehicle kinematic data, and physiological data as well as road alignment characteristics as input features for the proposed model that employs the long short-term memory (LSTM) network algorithm. Models with different combinations of observation window (lookback) and interval window (delay) were compared to derive the best window combination. The algorithms selected for comparison against the LSTM algorithm are random forest, XGBoost, and LightGBM. The results show that the proposed LSTM-based method can effectively predict dangerous driving behavior on sharp curves. The optimal window combination derived using the LSTM algorithm is lookback = 20 m and delay = 20 m. The prediction performance of the proposed model is significantly better than that of the other three compared algorithms, with F1-scores of 84.8% and 86.0% for the medium and high risk categories, respectively. In addition, the proposed LSTM-based model that fuses multiple data sources is proven to outperform the model that uses only vehicle kinematics data. The dynamic prediction method proposed in this paper can contribute to the development of a real-time prediction and warning system for driving risks at vehicle terminals in the intelligent connected vehicle environment. [ABSTRACT FROM AUTHOR]

• Norway has adopted the Safe System approach to road safety management. • This is associated with a larger decline in killed or seriously injured road users. • The use of effective road safety measures has increased. • Causal relationship cannot be established. • Replication of the study in a different country is encouraged. The objective of this paper is to evaluate the effectiveness of the Safe System approach to road safety management, as implemented in Norway. The paper proposes simple operational definitions of key elements of the Safe System approach to road safety management. The relationship between these elements and changes over time in the number of killed or seriously injured road users in Norway is studied by means of negative binomial regression models. These models do not support a causal interpretation of the findings, but predict systematic patterns in findings that, if replicated in other data sets, at least make a causal interpretation plausible, although not incontestable. The findings reported in this paper are broadly consistent with theoretical predictions and therefore support the effectiveness of the Safe System approach. It is highly likely that the adoption of the Safe System approach to road safety management in Norway has contributed to a larger improvement in road safety than would otherwise have occurred. [ABSTRACT FROM AUTHOR]