Using latent class analysis and mixed logit model to explore risk factors on driver injury severity in single-vehicle crashes

Quantifying the heterogeneity impact of risk factors on regional bicycle crash frequency: A hybrid approach of clustering and random parameter model

The existence of internal and external heterogeneity has been established by numerous studies across various fields, including transportation and safety analysis. The findings from these studies underscore the complexity of crash data and the multifaceted nature of risk factors involved in accidents. However, most studies consider the effects of unobserved heterogeneity from one perspective -- either within clusters (internal) or between clusters (external) -- and do not investigate the biases from both simultaneously on crash frequency analysis. To fill this gap, this study proposes a hybrid approach combining latent class cluster analysis with the random parameter negative binomial regression model (LCA-RPNB) to explore the association between risk factors and bicycle crash frequency. First, the bicycle crash data is categorized into three clusters using LCA based on crash features such as gender, trip purposes, weather, and light conditions. Then, the separated crash frequency models for different clusters and the overall model are developed based on RPNB using regional factors of crash locations as independent variables and the crash frequency of different clusters respectively as dependent variables. The hybrid approach enables a comprehensive examination of internal and external heterogeneities among bicycle crash frequency factors simultaneously. Results suggest that the proposed hybrid approach exhibits superior fitting and predictive performance compared to the model only considers the effects of unobserved heterogeneity from one perspective with the lower values of Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). This approach can help policymakers and urban planners to design more effective safety interventions by understanding the distinct needs of different bicyclist clusters and the specific factors that contribute to crash risk in each group.

Age-related driver injury occurrence from crashes at curve-grade combined segments

OBJECTIVES

Due to their relatively complex roadway characteristics, horizontal and vertical curve segments are associated with decreased visibility and a higher risk of rollovers. Multiple studies have identified the associated risk of young and older drivers separately in such complicated driving environments. This study investigated the relationship between driver age and injury occurrence from crashes occurring along curve-grade combined segments.

METHODS

Crash data recorded in Ohio State between 2012 and 2017 were used in this study. Driver age was categorized into 3 groups: teen (age <20 years), adult (age 20-64), and older adult (age >64). Descriptive statistics were summarized using random forest, gradient boosting, and extreme gradient boosting (XGBoost) to estimate the probability of a driver incurring an injury in case of a crash at curve-grade combined segments. The area under the receiver operating characteristics curve (AUROC) was used to select the best performing model. Partial dependence plots (PDPs) were used to interpret the model results.

RESULTS

The probability of injury occurrence is different for older drivers compared to teen and adult drivers. Although teen and adult drivers showed a higher probability of sustaining injuries in crashes with an increase in the degree of curvature, older drivers were more likely to sustain injuries in roadways with higher annual average daily traffic (AADT), steeper grades, and more occupants in the vehicle. Older drivers were observed to have a higher probability of sustaining injuries during peak hours and when unrestrained compared to teen and adult drivers.

CONCLUSIONS

The results emphasize the significance of tailored education and outreach countermeasures, particularly for teen and older drivers, aimed at decreasing the likelihood of injuries in such driving environments. This research adds to the expanding body of knowledge concerning the age-related occurrence of driver injuries resulting from crashes at curve-grade combined segments. The study findings provide insights into the potential over- or underrepresentation of certain age groups in analyzing crash injury occurrence. The insights gained from the machine learning analysis could also assist policymakers, transportation agencies, and traffic safety experts in developing targeted strategies to enhance road safety and protect vulnerable age groups.

Interpretable machine learning for evaluating risk factors of freeway crash severity

Machine learning (ML) models are widely employed for crash severity modelling, yet their interpretability remains underexplored. Interpretation is crucial for comprehending ML results and aiding informed decision-making. This study aims to implement an interpretable ML to visualize the impacts of factors on crash severity using 5 years of freeways data from Iran. Methods including classification and regression trees (CART), K-nearest neighbours (KNNs), random forest (RF), artificial neural network (ANN) and support vector machines (SVM) were applied, with RF demonstrating superior accuracy, recall, F1-score and ROC. The accumulated local effects (ALE) were utilized for interpretation. Findings suggest that light traffic conditions (volume / capacity < 0.5 ) with critical values around 0.05 or 0.38, and higher proportion of large trucks and buses, particularly at 10% and 4%, are associated with severe crashes. Additionally, speeds exceeding 90 km/h, drivers younger than 30 years, rollover crashes, collisions with fixed objects and barriers, nighttime driving and driver fatigue elevate the likelihood of severe crashes.

Analyzing the transition from two-vehicle collisions to chain reaction crashes: A hybrid approach using random parameters logit model, interpretable machine learning, and clustering

Chain reaction crashes (CRC) begin with a two-vehicle collision and rapidly intensify as more vehicles get directly involved. CRCs result in more extensive damage compared to two-vehicle crashes and understanding the progression of a two-vehicle collision into a CRC can unveil preventive strategies that have received less attention. In this study, to align with recent research direction and overcome the limitations of econometric and machine learning (ML) modelling, a hybrid approach is adopted. Moreover, to tackle the existing challenges in crash analysis, addressing unobserved heterogeneity in ML, and exploring random parameter effects and interactions more precisely, a new approach is proposed. To achieve this, a hybrid random parameter logit model and interpretable ML, joint with prior latent class clustering is implemented. Notably, this is the first attempt at using a clustering with hybrid modeling. The significant risk factors, their critical values, distinct effects, and interactions are interpreted using both marginal effects and the SHAP (SHapley Additive exPlanations) method across clusters. This study utilizes crash, traffic, and geometric data from eleven suburban freeways in Iran collected over a 5-year period. The overall results indicate an increased risk of CRC in congested traffic, higher traffic variation, and on horizontal curves combined with longitudinal slopes. Some parameters exhibit distinct or fluctuating effects, which are discussed across different conditions or considering interactions. For instance, during nighttime, heightened congestion on 2-lane freeways, increased traffic variation in less congested conditions, and adverse weather combined with horizontal curves and slopes pose risks. During daytime, increased traffic variation within highly congested sections, higher proportion of heavy vehicle traffic in moderately congested sections, and two lanes in each direction coupled with curves, elevate the levels of risk. The results of this study provide a better understanding of risk factors impact across different conditions, which are usable for policy makers.

Examining factors affecting driver injury severity in speeding-related crashes: a comparative study across driver age groups

This paper investigates the factors influencing the severity of driver injuries in single-vehicle speeding-related crashes, by comparing different driver age groups. This study employed a random threshold random parameter hierarchical ordered probit model and analysed crash data from Thailand between 2012 and 2017. The findings showed that young drivers face a heightened fatality risk when speeding in passenger cars or pickup trucks, hinting at the role of inexperience and risk-taking behaviours. Old drivers exhibit an increased fatality risk when speeding, especially in rainy conditions, on flush median roads, and during evening peak hours, attributed to reduced reaction times and vulnerability to adverse weather. Both young and elderly drivers face escalated fatality risks when speeding on road segments lacking guardrails during adverse weather, with older drivers being particularly vulnerable in rainy conditions. All age groups show an elevated fatality risk when speeding on barrier median roads, underscoring the significant role of speeding, which increases crash impact and limits margins of error and manoeuvrability, thereby highlighting the need for safety measures focusing on driver behaviour. These findings underscore the critical imperative for interventions addressing not only driver conduct but also road infrastructure, collectively striving to curtail the severity of speeding-related crashes.

Analysis of factors influencing the degree of accidental injury of bicycle riders considering data heterogeneity and imbalance

Bicycle safety has emerged as a pressing concern within the vulnerable transportation community. Numerous studies have been conducted to identify the significant factors that contribute to the severity of cyclist injuries, yet the findings have been subject to uncertainty due to unobserved heterogeneity and class imbalance. This research aims to address these issues by developing a model to examine the impact of key factors on cyclist injury severity, accounting for data heterogeneity and imbalance. To incorporate unobserved heterogeneity, a total of 3,895 bicycle accidents were categorized into three homogeneous sub-accident clusters using Latent Class Cluster Analysis (LCA). Additionally, five over-sampling techniques were employed to mitigate the effects of data imbalance in each accident cluster category. Subsequently, Bayesian Network (BN) structure learning algorithms were utilized to construct 32 BN models after pairing the accident data from the four accident cluster types before and after sampling. The optimal BN models for each accident cluster type provided insights into the key factors associated with cyclist injury severity. The results indicate that the key factors influencing serious cyclist injuries vary heterogeneously across different accident clusters. Female cyclists, adverse weather conditions such as rain and snow, and off-peak periods were identified as key factors in several subclasses of accident clusters. Conversely, factors such as the week of the accident, characteristics of the trafficway, the season, drivers failing to yield to the right-of-way, distracted cyclists, and years of driving experience were found to be key factors in only one subcluster of accident clusters. Additionally, factors such as the time of the crash, gender of the cyclist, and weather conditions exhibit varying levels of heterogeneity across different accident clusters, and in some cases, exhibit opposing effects.

Efficient and robust estimation of single-vehicle crash severity: A mixed logit model with heterogeneity in means and variances

This study delves into the factors that contribute to the severity of single-vehicle crashes, focusing on enhancing both computational speed and model robustness. Utilizing a mixed logit model with heterogeneity in means and variances, we offer a comprehensive understanding of the complexities surrounding crash severity. The analysis is grounded in a dataset of 39,788 crash records from the UK's STATS19 database, which includes variables such as road type, speed limits, and lighting conditions. A comparative evaluation of estimation methods, including pseudo-random, Halton, and scrambled and randomized Halton sequences, demonstrates the superior performance of the latter. Specifically, our estimation approach excels in goodness-of-fit, as measured by ρ2, and in minimizing the Akaike Information Criterion (AIC), all while optimizing computational resources like run time and memory usage. This strategic efficiency enables more thorough and credible analyses, rendering our model a robust tool for understanding crash severity. Policymakers and researchers will find this study valuable for crafting data-driven interventions aimed at reducing road crash severity.

Injury severity analysis of single-vehicle and two-vehicle crashes with electric scooters: A random parameters approach with heterogeneity in means and variances

In recent years, the electric scooter has become one of the most popular means of transportation on short trips. Due to the lag in the formulation of transportation policies and regulations, coupled with the increasing number of electric scooter crashes, there has been growing concern about the safety of pedestrians and electric scooter riders. For the first time in the extant literature, this study aims to analyze injury severity of electric scooter crashes by unobserved heterogeneity modeling approaches. A random parameters approach with heterogeneity in means and variances is utilized to examine the factors influencing injury severity, using data collected from the STATS19 road safety database. Electric scooter crashes are classified as single-vehicle crashes and two-vehicle crashes, with injury severity categorized into two groups: fatalities or serious injuries, and slight injuries. The model estimation was conducted by considering several variables including roadway, environment, temporality, vehicle, and rider characteristics, as well as second-party vehicle and driver characteristics and manners of collision specific to two-vehicle crashes. The results of the model estimation reveal that certain factors had relatively stable effects with the varying degree of crash injury severity outcomes in both single-vehicle crashes and two-vehicle crashes. These factors include nighttime incidents, weekdays, male riders, and an increase in rider age, all of which are associated with more severe injury outcomes. Moreover, the random parameters logit model with heterogeneity in means and variances is more flexible in accounting for unobserved heterogeneity and exhibits better goodness of fit. This study improves the understanding of electric scooter safety, and the finding can better inform public policy regarding electric scooter use to improve road safety and reduce injury severity of electric scooter crashes.

Analyzing driver injury severity in two-vehicle rear-end crashes considering leading-following configurations based on passenger car and light truck involvement

Rear-end crash is a major type of traffic crashes leading to a large number of injuries and fatalities each year, and passenger cars and light trucks are two main vehicle types in rear-end crashes on US roadways. Passenger cars and light trucks are different in size, vehicle mass and driver's vision. It is necessary to investigate the driver injury outcome patterns in rear-end crashes between passenger cars and light trucks considering crash configurations regarding the leading and following vehicle types. This study employs latent class multinomial logit (MNL) model to examine the risk factors on driver injury severity along with heterogeneity in variable effects presented by the cluster pattern in two-vehicle rear-end crashes involving passenger cars and light trucks, considering four crash configuration types, i.e., a passenger car struck by a passenger car, a light truck struck by a light truck, a passenger car struck by a light truck, and a light truck struck by a passenger car as exploratory variables. A model with two latent classes, which indicates the heterogeneity in variable effects among all the observations, is found to best fit the 7-year crash dataset from Washington State. The pseudo-elasticities are calculated to quantify the marginal effects of the contributing factors. The risk factors curve and sloping road condition, driver without seatbelt, and driver age of 65 and above increase driver fatality and serious injury risk greatly, and these three factors contribute from different latent classes. The crash configuration of a passenger car struck by a light truck is found to be one of class characteristics factors, which indicates that the heterogeneity exists between these two vehicle types. This factor is also a risk factor of injury. Furthermore, the leading vehicle is found to be much more vulnerable and closely related to injury, especially when it is in the crash of a passenger car struck by a light truck. The latent classes discovered give theoretical evidence of how to appropriately select subset data for further model construction for practical interest of serious injury prevention. The risk factors and their influence on injury severity provide beneficial insights on developing relevant countermeasures and strategies for injury severity mitigation on rear-end crashes involving passenger cars and light trucks.

Exploring the differences in rider injury severity in vehicle-two-wheelers accidents with dissimilar fault parties

Objective: The division of responsibility in vehicle-two-wheelers accidents reflects the extent to which different fault parties contributed to the occurrence of the accident, with significant differences in the injuries sustained by the riders in accidents where diverse parties were primarily responsible. We want to explore the difference in the severity of injury of riders in different fault parties of accidents so that we can make targeted protection improvements.Methods: In this study, three generalized ordered logit models were established for the total sample (n = 1204), the sample with drivers as the primary fault party (n = 607), and the sample with riders as the primary fault party (n = 597), respectively, to explore the differential impact factors on rider injury severity in vehicle-two-wheelers accidents involving different fault parties. Inter-group difference tests were conducted on the mean rider injury severity caused by differential factors in different accidents. Combining the impact effect trends and mean differences in the model, the differences in rider injury severity in accidents involving different fault parties were analyzed from the standpoints of human, vehicle, and road factors.Results: It was found that the effects of curve on injury severity was sheerly opposite in accidents with different fault parties and that factors, such as visual obstruction, road surface condition, gender, and helmet wearing differed in their effects on rider injury severity under different fault parties accidents. This reveals the driving tendencies and states of both parties in different environments.Conclusion: Based on the differential impact factor analysis and rider injury characteristics in accidents involving different fault parties, suggestions for improvement were made from the perspectives of road facilities, and safety awareness of drivers and riders, which are beneficial for improving rider safety and providing a theoretical reference for future regulations on liability allocation.

A hybrid clustering and random forest model to analyse vulnerable road user to motor vehicle (VRU-MV) crashes

The main goal of this study is to investigate the unobserved heterogeneity in VRU-MV crash data and to determine the relatively important contributing factors of injury severity. For this end, a latent class analysis (LCA) coupled with random parameters logit model (LCA-RPL) is developed to segment the VRU-MV crashes into relatively homogeneous clusters and to explore the differences among clusters. The random-forest-based SHapley Additive exPlanation (RF-SHAP) approach is used to explore the relative importance of the contributing factors for injury severity in each cluster. The results show that, vulnerable group (VG), intersection or not (ION) and road type (RT) clearly distinguish the crash clusters. Moto-vehicle type and functional zone have significant impact on the injury severity among all clusters. Several variables (e.g. ION, crash type [CT], season and RT) demonstrate a significant effect in a specific sub-cluster model. Results of this study provide specific and insightful countermeasures that target the contributing factors in each cluster for mitigating VRU-MV crash injury severity.

Temporal instability and differences in injury severity between restrained and unrestrained drivers in speeding-related crashes

Upon detecting a crash impact, the vehicle restraint system locks the driver in place. However, external factors such as speeding, crash mechanisms, roadway attributes, vehicle type, and the surrounding environment typically contribute to the driver being jostled within the vehicle. As a result, it is crucial to model unrestrained and restrained drivers separately to reveal the true impact of the restraint system and other factors on driver injury severities. This paper aims to explore the differences in factors affecting injury severity for seatbelt-restrained and unrestrained drivers involved in speeding-related crashes while accounting for temporal instability in the investigation. Utilizing crash data from Thailand between 2012 and 2017, mixed logit models with heterogeneity in means and variances were employed to account for multi-layered unobserved heterogeneity. For restrained drivers, the risk of fatal or severe crashes was positively associated with factors such as male drivers, alcohol influence, flush/barrier median roadways, sloped roadways, vans, running off the roadway without roadside guardrails, and nighttime on unlit or lit roads. For unrestrained drivers, the likelihood of fatal or severe injuries increased in crashes involving older drivers, alcohol influence, raised or depressed median roadways, four-lane roadways, passenger cars, running off the roadway without roadside guardrails, and crashes occurring in rainy conditions. The out-of-sample prediction simulation results are particularly significant, as they show the maximum safety benefits achievable solely by using a vehicle's seatbelt system. Likelihood ratio test and predictive comparison findings highlight the considerable combined impact of temporal instability and the non-transferability of restrained and unrestrained driver injury severities across the periods studied. This finding also demonstrates a potential reduction in severe and fatal injury rates by simply replicating restrained driver conditions. The findings should be of value to policymakers, decision-makers, and highway engineers when developing potential countermeasures to improve driver safety and reduce the frequency of severe and fatal speeding-related single-vehicle crashes.

Mitigating the impact of outliers in traffic crash analysis: A robust Bayesian regression approach with application to tunnel crash data

Traffic crash datasets are often marred by the presence of anomalous data points, commonly referred to as outliers. These outliers can have a profound impact on the results obtained through the application of traditional methods such as logit and probit models, commonly used in the domain of traffic safety analysis, resulting in biased and unreliable estimates. To mitigate this issue, this study introduces a robust Bayesian regression approach, the robit model, which utilizes a heavy-tailed Student's t distribution to replace the link function of these thin-tailed distributions, effectively reducing the influence of outliers on the analysis. Furthermore, a sandwich algorithm based on data augmentation is proposed to enhance the estimation efficiency of posteriors. The proposed model is rigorously tested using a dataset of tunnel crashes, and the results demonstrate its efficiency, robustness, and superior performance compared to traditional methods. The study also reveals that several factors such as night and speeding have a significant impact on the injury severity of tunnel crashes. This research provides a comprehensive understanding of the outliers treatment methods in traffic safety studies and offers valuable recommendations for the development of appropriate countermeasures to effectively prevent severe injuries in tunnel crashes.

Analyzing the impact of curve and slope on multi-vehicle truck crash severity on mountainous freeways

Many studies examine the road characteristics that impact the severity of truck crash accidents. However, some only analyze the effect of curves or slopes separately, ignoring their combination. Therefore, there are nine types of the combination of curve and slope in this study. The combination of curve and slope factor that affected the injury severity of truck crashes on mountainous freeways was examined using a correlated random parameter logit model. This method is applied to evaluate the correlation between the random parameters and those that exhibit unobserved heterogeneity. Also, the multinomial logit model and traditional random parameter logit model are used. The study's data were collected from multi-vehicle truck crashes on mountainous freeways in China. The results showed that the correlated random parameters logit model was better than the others. In addition, they demonstrated a correlation between the random parameters. Based on the estimation coefficients and marginal effects, the combination of curve and slope has a great influence on the injury severity of truck crashes. The main finding is that curve with medium radius and medium slope will significantly increase the probability of medium severity comparing to curve with high radius and flat slope. On the other hand, the injury severity of truck accidents was significantly impacted by crash type, vehicle type, surface condition, time of day, season, lighting condition, pavement type, and guardrail. Variables such as sideswipe, head-on, medium trucks, morning, dawn or dusk and summertime reduced the probability of truck crashes. Rollover, winter, gravel, and guardrail variables increased the risk of truck crashes. Correlations were also discovered between a rollover and dry surface condition and rollover and gravel pavement type. The research findings will help traffic officials determine effective countermeasures to decrease the severity of truck crashes on mountainous freeways.

Understanding the effect of sociodemographic and psychological latent characteristics on flex-route transit acceptance

Flex-route transit (FRT) has significant advantages in low-demand areas. Existing studies have focused on practical experience, strategic planning, and operational planning. Few studies have addressed the effect of sociodemographic and psychological latent characteristics on the acceptance of FRT. This study aims at exploring the effect of sociodemographic and psychological latent characteristics on FRT acceptance. To finish the goal, a household survey is conducted from April to May 2020 in Nanjing, China. The survey includes sociodemographic characteristics and observed variables of individuals. Firstly, the study extracts six psychological latent characteristics to reflect individuals' attitudes based on previous and mature researches in the field of technology acceptance model (TAM) and theory of planned behavior (TPB). Then, a multiple indicators and multiple causes (MIMIC) is applied to calculate six psychological latent characteristics. Finally, an integrated model, consisting of the MIMIC and a binary logit model (BLM), is applied to match sociodemographic and psychological latent characteristics. The BLM with sociodemographic characteristics is developed as the reference model to compare the effects of psychological latent characteristics. Results show that psychological latent factors play a significant role in estimating the effect on FRT acceptance. Results of the integrated model show that the parameter of car is -0.325, displaying individuals with private cars are more reluctant to use FRT. Therefore, restricting private cars is an effective measure to facilitate FRT. Improving flexibility (0.241) is a significant measure to facilitate FRT. Findings are expected to facilitate decision-making of transport planners and engineers, and therefore enhance the service of the FRT system.

Crash characteristics for classic/historic vehicles and comparisons to newer vehicles

INTRODUCTION

Older vehicles, commonly referred to as "classic," "vintage," or "historic" vehicles (CVH), share the roadways with newer vehicles. Older vehicles lacking safety systems likely come with an increased risk of fatality, however there is no study examining the typical conditions for crashes involving CVH.

METHOD

This study utilized information from crashes occurring in 2012 to 2019 to estimate fatal crash rates for vehicles grouped by model year deciles. Data from crashes documented in the National Highway Traffic Safety Administration's (NHTSA) FARS and GES/CRSS data sets were utilized to examine roadway, temporal, and crash types for passenger vehicles produced in 1970 or earlier (CVH).

RESULTS

These data show CVH crashes are rare (<1% of crashes), but carry a relative risk of fatality from 6.70 (95th CI: 5.44-8.26) for impacts with other vehicles, which was the most common crash, to 9.53 (7.28-12.47) for rollovers. Most crashes occurred in dry weather, typically during summer, in rural areas, most frequently on two lane roads, and in areas with speed limits between 30 and 55 mph. Factors associated with fatality for occupants in CVH included alcohol use, lack of seat belt use, and older age.

CONCLUSIONS AND PRACTICAL APPLICATIONS

Crashes involving a CVH are a rare event but have catastrophic consequences when they do occur. Regulations that limit driving to daylight hours may lower the risk of crash involvement, and safety messaging to promote belt use and sober driving may also help. Additionally, as new "smart" vehicles are developed, engineers should keep in mind that older vehicles remain on the roadway. New driving technologies will need to safely interact with these older, less safe vehicles.

Relationship between Vehicle Safety Ratings and Drivers' Injury Severity in the Context of Gender Disparity

Previous studies have analyzed the relationship between vehicle safety ratings from impact tests and actual crash injury severity. Nevertheless, no study has investigated the relationship in the context of gender disparity. The main objective of this paper is to explore the validity of the 5-star ratings of the U.S. National Highway Traffic Safety Administration, which describes vehicles’ protectiveness, using actual traffic crash data by gender. Random parameter models are developed using 2015–2020 two-vehicle crash data from Maryland, United States. According to the data, over 90% of vehicles have 4–5 stars in overall, front-impact, and side-impact 5-star ratings. After controlling other factors, it is shown that woman drivers are more likely to be seriously injured in two-vehicle crashes than men drivers when using vehicles with the same 5-star safety ratings. Moreover, there is significant individual heterogeneity in the effect of vehicles with different 5-star safety ratings on driver injury severity. Using vehicles with more stars can reduce the risk of being seriously injured for most man drivers. However, the probability of woman drivers being seriously injured is reduced by approximately 5% on average by using vehicles with higher star ratings in the overall and front-impact 5-star rating, and individual heterogeneity shows a difference of nearly 50% in positive and negative effects. The overall and front-impact 5-star ratings of vehicles could not provide reasonable information as the safety performance of vehicles in traffic crashes for woman drivers. On the other hand, drivers’ residence, driving characteristics, crash types, and environmental characteristics are significantly associated with the injury severity. It is expected that the results from this study will contribute to guide a better vehicle safety design for both men and women.

Injury severity analysis of two-vehicle crashes at unsignalized intersections using mixed logit models

The severity of the two-vehicle crash is closely related to the characteristics of both the struck and striking vehicles. Ignoring vehicle roles may lead to biased results. Thus, this study used mixed logit models to determine the factors that influence injury severity in the two-vehicle crash, taking into account the vehicle characteristics of the different crash roles. The data used is collected from Pennsylvania Department of Transportation (PennDOT) Open Data Portal. First, the synthetic minority oversampling technique and nearest neighbors (SMOTE-ENN) strategy was selected to address the class imbalance problem of crash data. Then, two separated mixed logit models were developed for four- and three-legged unsignalized intersections. The results suggest that the type and movement of vehicles have significant effects on crash severity. For example, right-turn vehicles being struck can lead to more serious crashes than striking other vehicles. Large trucks striking other vehicles are found to increase crash severity, but being struck is found to decrease crash severity. Additionally, several factors were also identified to affect crash severity in both models and effective countermeasures suggestions were proposed to mitigate crash severity.Supplemental data for this article is available online at at .

Influence factors on injury severity of bicycle-motor vehicle crashes: A two-stage comparative analysis of urban and suburban areas in Beijing

OBJECTIVE

More attention should be given to bicycle-motor vehicle (BMV) crashes, as cyclists are at a higher risk of suffering injuries than motor vehicle users in a crash. This study aims to explore the factors influencing the injury severity of bicycle-motor vehicle (BMV) crashes in Beijing (China) and discusses the commonalities and differences between the urban and suburban areas.

METHODS

Information regarding 1,136 crashes between bicycles and motor vehicles were collected using police reported data from 2014 to 2015. A two-stage approach integrating random parameters logit (RP-logit) model and two-step clustering (TSC) algorithm was proposed to investigate the significant influence factors and their combination characteristics. Specifically, the RP-logit model was first used to identify the significant influence factors of urban and suburban areas, and then the TSC algorithm was applied to reveal the combination characteristics of significant influence factors for the fatal crashes.

RESULTS

Five factors were found to be statistically significant and had random effects on the injury severity in urban areas, i.e., type of motor vehicle, motor vehicle license ownership, type of bicycle, signal control mode and lighting condition; and seven factors were found to be statistically significant on the injury severity in suburban areas, i.e., type of motor vehicle, motor vehicle license ownership, physical isolation facility, signal control mode, weather, visibility and lighting condition. Based on TSC, the combination of significant factors showed different characteristics for fatal crashes in urban and suburban areas, in which two types of the scene including five factors should be concerned in urban areas while one type of scene containing four factors in suburban areas.

CONCLUSIONS

The results suggest that different influence factors and individual heterogeneity exist in the RP-logit model for injury severity analysis of BMV crashes in urban and suburban areas. It shows that in urban areas, heavy truck, light truck and bus significantly increase the likelihood of fatal injury than that of suburban areas. These findings can provide valuable reference information for BMV crashes response, such as heavy truck restriction, to facilitate regional safety measures for urban and suburban areas.

Classification and pattern extraction of incidents: a deep learning-based approach

Classifying or predicting occupational incidents using both structured and unstructured (text) data are an unexplored area of research. Unstructured texts, i.e., incident narratives are often unutilized or underutilized. Besides the explicit information, there exist a large amount of hidden information present in a dataset, which cannot be explored by the traditional machine learning (ML) algorithms. There is a scarcity of studies that reveal the use of deep neural networks (DNNs) in the domain of incident prediction, and its parameter optimization for achieving better prediction power. To address these issues, initially, key terms are extracted from the unstructured texts using LDA-based topic modeling. Then, these key terms are added with the predictor categories to form the feature vector, which is further processed for noise reduction and fed to the adaptive moment estimation (ADAM)-based DNN (i.e., ADNN) for classification, as ADAM is superior to GD, SGD, and RMSProp. To evaluate the effectiveness of our proposed method, a comparative study has been conducted using some state-of-the-arts on five benchmark datasets. Moreover, a case study of an integrated steel plant in India has been demonstrated for the validation of the proposed model. Experimental results reveal that ADNN produces superior performance than others in terms of accuracy. Therefore, the present study offers a robust methodological guide that enables us to handle the issues of unstructured data and hidden information for developing a predictive model.

Risk Factors Analysis of Car Door Crashes Based on Logistic Regression

Unlike door crash accidents predominantly involving bicycles in Australia, the UK, and other Western countries, cases in Taiwan are far more fatal as they usually involve motorcycles. This is due to the unique anthropogeography and transportation patterns of Taiwan, particularly the numbers of motorcycles being twice that of cars. Both path analysis and multivariate logistic regression methods were adopted in this study. The multivariate logistic regression analysis results have shown that the main risk factors causing serious injuries in door crashes include winter, morning, male motorcyclists, heavy motorcycles, and the left sides of cars. Regarding the gender differences in motorcyclists, it appears that female motorcyclists have higher door crash accident rates, while the odds of severe injury and fatality in male motorcyclists are 1.658 times greater than that of female motorcyclists. The risk factors derived from the multivariate logistic regression analysis were further discussed and analysed. It was found that the causes of serious injuries and deaths stemming from door crashes were related to the risk perception ability, reaction ability, visibility, and riding speed of the motorcyclists. Therefore, suggestions on risk management and accident prevention were proposed using advocacy through the 3E strategies of human factors engineering design.

Factors affecting injury severity and the number of vehicles involved in a freeway traffic accident: investigating their heterogeneous effects by facility type using a latent class approach

The number of vehicles involved in a traffic accident can be representative of the severity of the accident and provide profound insight into the diverse factors affecting severity, which cannot be identified through the victim fatality rate. This paper presents an analysis and comparison between the effects of factors affecting injury severity and the number of involved vehicles. In this study, a latent class model was used to investigate the unobserved heterogeneity of the accident factors. Freeway facility types are latent factors that affect the heterogeneity of the effects of accident factors. The class mainly including accidents at the freeway mainline sections included more injury/fatal accidents and multiple-vehicle accidents and more significant accident factor estimation results than the other class including accidents at the tollgates or ramps. Among these factors, night-time, faults made by the driver, and heavy vehicle accidents were found to increase the accident severity. Investigating accident factors affecting both the injury severity and number of involved vehicles is important as the number of people who are injured or dead is likely to increase when multiple vehicles are involved in the accident.

Exploring relationships between microscopic kinetic parameters of tires under normal driving conditions, road characteristics and accident types

INTRODUCTION

Freeway accidents are a leading cause of death in China, which also triggers substantial economic loss and an emotional burden to society. However, the internal mechanism of how microscopic kinetic parameters of vehicles influenced by road characteristics determine the occurrence of different types of accidents has not been explicitly studied. This research aimed to explore the "link role" of tire microscopic kinetic parameters in road characteristic variables and traffic accidents to aid in facilitating the traffic design and management, and thus to prevent traffic accident.

METHOD

A mountain freeway in Zhejiang Province, China was used as the research object and the data used in this paper were obtained through a real-time vehicle experiment. Multiple estimation models, including the standard ordered logit (SOL) model, fixed parameters logit (FPL) model, and random parameters logit (RPL) model were established.

RESULTS

The findings show that road characteristics will affect the longitudinal kinetic characteristics of the vehicle and, consequently, map the level of risk of rear-end accidents. Driving compensation effects were also identified in this paper (i.e., the drivers tend to be more cautious in complicated driving circumstances). Another finding relating to the mountain freeway is that different tunnel characteristics (e.g., tunnel entrance and tunnel exit) have different effects on different types of traffic accidents. Practical Applications: The framework proposed in this article can provide new insight for researchers to enlarge the research subjects of both explanatory and outcome variables in accident analysis. Future research could be implemented to consider more driving conditions.

Injury severity prediction of traffic crashes with ensemble machine learning techniques: a comparative study

A better understanding of injury severity risk factors is fundamental to improving crash prediction and effective implementation of appropriate mitigation strategies. Traditional statistical models widely used in this regard have predefined correlation and intrinsic assumptions, which, if flouted, may yield biased predictions. The present study investigates the possibility of using the eXtreme Gradient Boosting (XGBoost) model compared with few traditional machine learning algorithms (logistic regression, random forest, and decision tree) for crash injury severity analysis. The data used in this study was obtained from the traffic safety department, ministry of transport (MOT) at Riyadh, KSA, and contains 13,546 motor vehicle collisions along 15 rural highways reported between January 2017 to December 2019. Empirical results obtained using k-fold (k = 10) for various performance metrics showed that the XGBoost technique outperformed other models in terms of the collective predictive performance as well as injury severity individual class accuracies. XGBoost feature importance analysis indicated that collision type, weather status, road surface conditions, on-site damage type, lighting conditions, and vehicle type are the few sensitive variables in predicting the crash injury severity outcome. Finally, a comparative analysis of XGBoost based on different performance statistics showed that our model outperformed most previous studies.

Single-vehicle crash severity outcome prediction and determinant extraction using tree-based and other non-parametric models

Single-vehicle crashes are more fatality-concentrated and have posed increasing challenges in traffic safety, which is of great research necessity. Tremendous previous studies have conducted relevant analysis with econometric modeling approaches, whereas the ability of non-parametric methods to predict crash severity is still smattering of knowledge. Consequently, the main objective of this paper is to conduct single-vehicle crash severity prediction with different tree-based and non-parameter models. An alternate aim is to identify the intrinsic mechanism of how contributing factors determine single-vehicle crash severity. By virtue of Grid-Search method, this paper conducted fine-tuning of different models to obtain the best performances based on five crash severity sub-datasets. For model evaluation, the accuracy indicators were calculated in training, validation and test sets, respectively. Besides, feature importance extraction was undertaken based on the results of model comparison. The finding indicated that these models didn't exhibit a huge performance difference for crash severity prediction in the same severity level; however, the performances of the models did vary among different datasets, with an average training accuracy of 99.27 %, 96.4 %, 86.98 %, 86.84 %, 71.76 % in fatal injury, severe injury, visible injury, complaint of pain, PDO crash datasets, respectively. Additionally, it was found that in each severity dataset, the indicator urban freeways is a determinant factor that leads to the occurrence of crashes while rural freeways is more related to more severe crashes (i.e., fatal and severe crashes). This paper can provide valuable information for model selection and tuning in accident severity prediction. Future research could consider the influences that temporal instability of contributing features has on the model performances.

Modeling Wrong-way Driving (WWD) crash severity on arterials in Florida

Wrong-way Driving (WWD) is the movement of a vehicle in a direction opposite to the one designated for travel. WWD studies and mitigation strategies have exclusively been focused on limited-access facilities. However, it has been established that WWD crashes on arterial corridors are also severe and relatively more common. As such, this study focused on determining factors influencing the severity of WWD crashes on arterials. The analysis was based on five years of WWD crashes (2012-2016) that occurred on state-maintained arterial corridors in Florida. Police reports of 2,879 crashes flagged as "wrong-way" were downloaded and individually reviewed. The manual review of the police reports revealed that of the 2,879 flagged WWD crashes, only 1,890 (i.e., 65.6 %) occurred as a result of a vehicle traveling the wrong way. The Bayesian partial proportional odds (PPO) model was used to establish the relationship between the severity of these WWD crashes and different driver attributes, temporal factors, and roadway characteristics. The following variables were significant at the 90 % Bayesian Credible Interval (BCI): day of the week, lighting condition, presence of work zone, crash location, age and gender of the wrong-way driver, airbag deployment, alcohol use, posted speed limit, speed ratio (i.e., driver's speed over the posted speed limit), and the manner of collision. Based on the model results, specific countermeasures on Education, Engineering, Enforcement, and Emergency response are discussed. Potential Transportation Systems Management and Operations (TSM&O) strategies for WWD detection systems on arterials to minimize WWD frequency and severity are also proposed.

A novel approach to set driving simulator experiments based on traffic crash data

Several studies have often cited crash occurrences as a motivation to perform a driving simulator experiment and test driver behavior to understand their causal relations. However, decisions regarding the simulated scenario and participants' requirements do not often rely directly on traffic crash data. To fill the gap between simulation and real data, we have proposed a new framework based on Clustering Analysis (K-medoids) to support the definition of driving simulator experiments when the purpose is to investigate the driver behavior under real risky road conditions to improve road safety. The suggested approach was tested with data of three years of police records regarding loss-of-control crashes and information on three Brazilian rural highways' geometry and traffic volume. The results showed the good suitability of the method to compile the data's diversity into four clusters, representing and summarizing the crashes' main characteristics in the region of study. Drivers' attributes (age and gender) were initially intended to integrate the clustering analysis; however, due to the sample's homogeneity of these characteristics, they did not contribute to the cluster definition. Hence, they were used simply to identify the target population for all scenarios. Therefore, we concluded that driving simulator experiments could benefit from the new approach since it identifies scenarios characterized by many variables connected to real risky situations and orients participants' recruitment leading to efficient safety analysis.

Hierarchical binary logit model to compare driver injury severity in single-vehicle crash based on age-groups

Most of the previous single-vehicle crash analysis studies ignored the effect of road-segments level at higher plan that could probably be unobserved heterogeneity and vary among crash-level factor from one road-segment to next and possibly could lead to a potential biased estimated result. This study developed a hierarchical binary logit model which have the ability to account for both unobserved heterogeneity and correlation within road-segment, to investigate and compare the impact of significant factors influencing fatal single-vehicle crash between young, mid-age and old driver model. A seven-years from 2011 to 2017 crash data, Department of Highway (DOH), Thailand were used in this study. The Intra-Class-Correlation values indicate the importance of road-segment level that 10.1%, 12.2% and 12.8% of the total variation were accounted by random effect from road-segment heterogeneity for young, mid-age and old driver model, respectively. The estimated result of this study shows that influence of alcohol and fatigue increase risk of fatal crash among young and old driver, seatbelt-usage reduce risk of being fatal among mid-age and old driver, roadside safety feature (guardrail) significantly reduce fatality risk among young and mid-age driver, and night time driving without light increase probability of fatal crash for mid-age driver. This study recommends the need to enforce the law on driver under influence of alcohol and seatbelt usage, educational campaign on driving, and installation of guardrail on curve road.

Assessing the explanatory and predictive performance of a random parameters count model with heterogeneity in means and variances

Random parameters model has been demonstrated to be an effective method to account for unobserved heterogeneity that commonly exists in highway crash data. However, the predefined single distribution for each random parameter may limit how the unobserved heterogeneity is captured. A more flexible approach is to develop a random parameters model with heterogeneity in means and variances by allowing the mean and variance of potential each random parameter to be an estimable function of explanatory variables. This burgeoning technique for modelling unobserved heterogeneity has been increasingly applied to various safety evaluation scenarios recently. However, the predictive performance of this emerging method, which determines the practicability of the model for a specific circumstance, has never been investigated as far as our knowledge. In addition, the explanatory power by including heterogeneous means and variances of random parameters need to be further investigated to confirm the potential merits of this method in crash data analysis. In this paper, a random parameters negative binomial with heterogeneity in means and variances (RPNBHMV) model, a standard random parameters negative binomial (RPNB) model and a traditional fixed parameters negative binomial (NB) model were estimated using the same dataset. The explanatory and predictive performance of the three models were thoroughly evaluated and compared. Results showed that: 1) the RPNB model fitted the data significantly better than the NB model, and the RPNBHMV model further improved the statistical fit of the RPNB model but the improvement was slight; 2) more insights into interactions of safety factors were inferred from the RPNBHMV model, which demonstrates the explanatory benefit of this model; 3) the RPNBHMV and RPNB models had both advantages (e.g., produced overall better prediction accuracy) and disadvantages (e.g., provided reduced prediction accuracy across the range of explanatory variables) when applied to in-sample observations (i.e., observations used to estimate the model); 4) the RPNBHMV and RPNB models might be less precise than the NB model when applied to out-of-sample observations. These findings indicate that the RPNBHMV model offers more insights and may be used for explanatory safety analysis for sites where reliable data can be collected. However, the simple NB model is more reliable - at least with the dataset used in this study - than its random parameters model counterparts for other sites where the data are unavailable or unreliable, which is a common safety evaluation scenario in practice.

Injury severity analysis of familiar drivers and unfamiliar drivers in single-vehicle crashes on the mountainous highways

Mountainous highways suffer from high crash rates and fatality rates in many countries, and single-vehicle crashes are overrepresented along mountainous highways. Route familiarity has been found greatly associated with driver behaviour and traffic safety. This study aimed to investigate and compare the contributory factors that significantly influence the injury severities of the familiar drivers and unfamiliar drivers involved in mountainous highway single-vehicle crashes. Based on 3037 cases of mountainous highway single-vehicle crashes from 2015 to 2017, the characteristics related to crash, environment, vehicle and driver are included. Random-effects generalized ordered probit (REGOP) models were applied to model injury severities of familiar drivers and unfamiliar drivers that are involved in the single-vehicle crashes on the mountainous highways, given that the single-vehicle crashes had occurred. The results of REGOP models showed that 8 of the studied factors are found to be significantly associated with the injury severities of the familiar drivers, and 10 of the studied factors are found to significantly influence the injury severities of unfamiliar drivers. These research results suggest that there is a large difference of significant factors contributing to the injury severities between familiar drivers and unfamiliar drivers. The results shed light on both the similar and different causes of high injury severities for familiar and unfamiliar drivers involved in mountainous highway single-vehicle crashes. These research results can help develop effective countermeasures and proper policies for familiar drivers and unfamiliar drivers targetedly on the mountainous highways and alleviate injury severities of mountainous highway single-vehicle crashes to some extent. Based on the results of this study, some potential countermeasures can be proposed to minimize the risk of single-vehicle crashes on different mountainous highways, including tourism highways with a large number of unfamiliar drivers and other normal mountainous highways with more familiar drivers.

Identifying heterogeneous factors for driver injury severity variations in snow-related rural single-vehicle crashes

Snowy weather is consistently considered as a hazardous factor due to its potential leading to severe fatal crashes. A seven-year crash dataset including rural highway single vehicle crashes from 2010 to 2016 in Washington State is applied in the present study. Pseudo elasticity analysis is conducted to investigate significant impact factors and the temporal stability of model specifications is tested via a likelihood ratio test. The proposed model based on the seven-year dataset is able to capture the individual-specific heterogeneity across crash records for four significant factors, i.e., surface ice, male, and airbag combine deployment for minor injury, and male for serious injury and fatality. Their estimated parameters were found to be normal distribution instead of fixed value over the observations. Other significant impact factors with fixed effects are: inroad object, animal, overturn, surface wet, surface snow, unusual horizontal design, medium and high speed limits, driver age, impaired condition, no belt usage, vehicle type, airbag deployment. Especially, when compared to significant factors for crashes under other weather conditions, male indicator and impaired condition show significant higher effects in snow-related crashes. The results of temporal stability test show that the model specification is generally not temporally stable for driver injury severity model based on the years of crash data that were used, especially for longer period (more than 3-year dataset). Models that allow the explanatory variables to track temporal heterogeneity, are of great interest and can be explored in future research.

Severity of passenger injuries on public buses: A comparative analysis of collision injuries and non-collision injuries

INTRODUCTION

Although public buses have been demonstrated as a relatively safe mode of transport, the number of injuries to public bus passengers is far from negligible. Existing studies of public bus safety have focused primarily on injuries caused by collisions. Surprisingly, limited effort has been devoted to identifying factors that increase the severity of passenger injuries in non-collision incidents.

METHOD

Our study therefore investigated the injury risk of public bus passengers involved in collision incidents and non-collision incidents comparatively, based on a police-reported dataset of 17,383 passengers injured on franchised public buses over a 10-year period in Hong Kong. A random parameters logistic model was established to estimate the likelihood of fatal and severe injuries to passengers as a function of various factors.

RESULTS

Our results indicated substantial inconsistences in the effects of risk factors between models of non-collision injuries and collision injuries. The severity of passenger injuries tended to increase significantly when non-collision incidents occurred due to excessive speed of bus drivers, on double-decker buses, in less urbanized areas, in winter, in heavy rains, during daytime, and at night without street lighting. Elderly female passengers were also found more likely to be fatally or severely injured in non-collision incidents if they lost their balance while boarding, alighting from, or standing on a bus. In comparison, the following factors were associated with a greater likelihood of fatal or severe injuries in collision incidents: elderly female passengers, standing passengers who lost balance, buses out of driver control, double-decker buses, collisions with vehicles or objects, and less urbanized areas. Practical Applications: Based on our comparative analysis, more targeted countermeasures, namely "4E" (engineering, enforcement, emergency, and education) and "3A" (awareness, appreciation, and assistance), were recommended to mitigate collision injuries and non-collision injuries to public bus passengers, respectively.

Applying hierarchical logistic models to compare urban and rural roadway modeling of severity of rear-end vehicular crashes

A rear-end crash is a widely studied type of road accident. The road area at the crash scene is a factor that significantly affects the crash severity from rear-end collisions. These road areas may be classified as urban or rural and evince obvious differences such as speed limits, number of intersections, vehicle types, etc. However, no study comparing rear-end crashes occurring in urban and rural areas has yet been conducted. Therefore, the present investigation focused on the comparison of diverse factors affecting the likelihood of rear-end crash severities in the two types of roadways. Additionally, hierarchical logistic models grounded in a spatial basis concept were applied by determining varying parameter estimations with regard to road segments. Additionally, the study compared coefficients with multilevel correlation model and those without multilevel correlation. Four models were established as a result. The data used for the study pertained to rear-end crashes occurring on Thai highways between 2011 and 2015. The results of the data analysis revealed that the model parameters for both urban and rural areas are in the same direction with the larger number of significant parameter values present in the rural rear-end crash model. The significant variables in both the urban and rural road segment models are the seat belt use, and the time of the incident. To conclude, the present study is useful because it provides another perspective of rear-end crashes to encourage policy makers to apply decisions that favor rules that assure safety.