Hybrid feature selection-based machine learning Classification system for the prediction of injury severity in single and multiple-vehicle accidents.

To undertake a reliable analysis of injury severity in road traffic accidents, a complete understanding of important attributes is essential. As a result of the shift from traditional statistical parametric procedures to computer-aided methods, machine learning approaches have become an important aspect in predicting the severity of road traffic injuries. The paper presents a hybrid feature selection-based machine learning classification approach for detecting significant attributes and predicting injury severity in single and multiple-vehicle accidents. To begin, we employed a Random Forests (RF) classifier in conjunction with an intrinsic wrapper-based feature selection approach called the Boruta Algorithm (BA) to find the relevant important attributes that determine injury severity. The influential attributes were then fed into a set of four classifiers to accurately predict injury severity (Naive Bayes (NB), K-Nearest Neighbor (K-NN), Binary Logistic Regression (BLR), and Extreme Gradient Boosting (XGBoost)). According to BA's experimental investigation, the vehicle type was the most influential factor, followed by the month of the year, the driver's age, and the alignment of the road segment. The driver's gender, the presence of a median, and the presence of a shoulder were all found to be unimportant. According to classifier performance measures, XGBoost surpasses the other classifiers in terms of prediction performance. Using the specified attributes, the accuracy, Cohen's Kappa, F1-Measure, and AUC-ROC values of the XGBoost were 82.10%, 0.607, 0.776, and 0.880 for single vehicle accidents and 79.52%, 0.569, 0.752, and 0.86 for multiple-vehicle accidents, respectively.

Driving specialist's ratings of on-road performance and naturalistic driving crashes and near-crashes.

BACKGROUND: This study aimed to evaluate the association between a Certified Driving Rehabilitation Specialist's (CDRS) ratings of on-road driving performance by older drivers and at-fault crash and near-crash involvement using naturalistic driving techniques where crashes and near-crashes are recorded in everyday driving through in-vehicle instrumentation. METHODS: This is a cohort study of 144 drivers aged 70 years and over who were recruited due to a recent ophthalmology clinic visit at the University of Alabama at Birmingham. Baseline measurements consisted of demographics, visual status, and other health variables. At-fault crashes and near-crashes over 6 months were identified through instrumentation placed in their personal vehicle that recorded vehicle kinematics and video. After 6 months, a CDRS completed an on-road assessment and provided a composite rating on specific driving behaviors and a global score. RESULTS: Rate ratios examining the association between older drivers with worse CDRS composite scores and rates of at-fault crashes, at-fault near-crashes, and combined at-fault crashes and near-crashes were significantly higher compared to drivers with better scores. Results were similar for the CDRS global score. CONCLUSIONS: Motor vehicle administrations use CDRS ratings to make decisions about licensure, and in clinical programs such as those based at rehabilitation clinics use them to make recommendations about fitness to drive and rehabilitation. This study suggests that these decisions and recommendations are valid from a safety standpoint.

Investigating the uniqueness of crash injury severity in freeway tunnels: A comparative study in Guizhou, China.

INTRODUCTION: With the rapid development of transportation infrastructures in precipitous areas, the mileage of freeway tunnels in China has been mounting during the past decade. Provided the semi-constrained space and the monotonous driving environment of freeway tunnels, safety concerns still remain. This study aims to investigate the uniqueness of the relationships between crash severity in freeway tunnels and various contributory factors. METHOD: The information of 10,081 crashes in the entire freeway network of Guizhou Province, China in 2018 is adopted, from which a subset of 591 crashes in tunnels is extracted. To address spatial variations across various road segments, a two-level binary logistic approach is applied to model crash severity in freeway tunnels. A similar model is also established for crash severity on general freeways as a benchmark. RESULTS: The uniqueness of crash severity in tunnels mainly includes three aspects: (a) the road-segment-level effects are quantifiable with the environmental factors for crash severity in tunnels, but only exist in the random effects for general freeways; (b) tunnel has a significantly higher propensity to cause severe injury in a crash than other locations of a freeway; and (c) different influential factors and levels of contributions are found to crash severity in tunnels compared with on general freeways. Factors including speed limit, tunnel length, truck involvement, rear-end crash, rainy and foggy weather and sequential crash have positive contributions to crash severity in freeway tunnels. Practical applications: Policy implications for traffic control and management are advised to improve traffic safety level in freeway tunnels.

Injury-severity analysis of lane change crashes involving commercial motor vehicles on interstate highways.

INTRODUCTION: One of the challenging tasks for drivers is the ability to change lanes around large commercial motor vehicles. Lane changing is often characterized by speed, and crashes that occur due to unsafe lane changes can have serious consequences. Considering the economic importance of commercial trucks, ensuring the safety, security, and resilience of freight transportation is of paramount concern to the United States Department of Transportation and other stakeholders. METHOD: In this study, a mixed (random parameters) logit model was developed to better understand the relationship between crash factors and associated injury severities of commercial vehicle crashes involving lane change on interstate highways. The study was based on 2009-2016 crash data from Alabama. RESULTS: Preliminary data analysis showed that about 4% of the observed crashes were major injury crashes and drivers of commercial motor vehicles were at-fault in more than half of the crashes. Acknowledging potential crash data limitations, the model estimation results reveal that there is increased probability of major injury when lane change crashes occurred on dark unlit portions of interstates and involve older drivers, at-fault commercial vehicle drivers, and female drivers. The results further show that lane change crashes that occurred on interstates with higher number of travel lanes were less likely to have major injury outcomes. Practical Applications: These findings can help policy makers and state transportation agencies increase awareness on the hazards of changing lanes in the immediate vicinity and driving in the blind spots of large commercial motor vehicles. Additionally, law enforcement efforts may be intensified during times and locations of increased unsafe lane changing activities. These findings may also be useful in commercial vehicle driver training and driver licensing programs.

Improving knowledge of cyclist crashes based on hospital data including crash descriptions from open text fields.

INTRODUCTION: In this study we explore the added value of bicycle crash descriptions from open text fields in hospital records from the Aarhus municipality in Denmark. We also explore how bicycle crash data from the hospital complements crash data registered by the police in the same area and time period. METHOD: The study includes 5,313 Danish bicycle crashes, of which 4,205 were registered at the hospital and 1,078 by the police. All crashes occurred from 2010 to 2015. We performed an in-depth analysis of the open text fields on hospital records to identify factors associated with each crash using four categories: bicyclist, road, bicycle, and the other party. We employed the chi-squared test to compare the distribution of variables between crashes registered at the hospital and by the police. A binary logit model was used to estimate the probability that a crash factor is identified, and that each crash factor is associated with a single-bicycle crash. RESULTS: The open-ended text fields in hospital records provide detailed information about crash factors not available in police records, including riding speed, inattention, clothing, specific road conditions, and bicycle defects. The factors alcohol and curb had the highest odds of being identified in relation to a single-bicycle crash. Crash data registered at the hospital included a larger number of bicycle crashes, particularly single-bicycle crashes and crashes with slight injuries only. CONCLUSION: Crash information registered at the hospital in Aarhus Municipality contributes to a better understanding of bicycle crashes due to detailed information about crash-associated factors as well as information about a larger number of bicycle crashes, particularly single-bicycle crashes. Practical implication: Efforts to improve access to detailed information about bicycle crashes are needed to provide a better basis for bicycle crash prevention.

Classification of road traffic injury collision characteristics using text mining analysis: Implications for road injury prevention.

Road traffic injuries are a leading cause of morbidity and mortality globally. Understanding circumstances leading to road traffic injury is crucial to improve road safety, and implement countermeasures to reduce the incidence and severity of road trauma. We aimed to characterise crash characteristics of road traffic collisions in Victoria, Australia, and to examine the relationship between crash characteristics and fault attribution. Data were extracted from the Victorian State Trauma Registry for motor vehicle drivers, motorcyclists, pedal cyclists and pedestrians with a no-fault compensation claim, aged > = 16 years and injured 2010-2016. People with intentional injury, serious head injury, no compensation claim/missing injury event description or who died < = 12-months post-injury were excluded, resulting in a sample of 2,486. Text mining of the injury event using QDA Miner and Wordstat was used to classify crash circumstances for each road user group. Crashes in which no other was at fault included circumstances involving lost control or avoiding a hazard, mechanical failure or medical conditions. Collisions in which another was predominantly at fault occurred at intersections with another vehicle entering from an adjacent direction, and head-on collisions. Crashes with higher prevalence of unknown fault included multi-vehicle collisions, pedal cyclists injured in rear-end collisions, and pedestrians hit while crossing the road or navigating slow traffic areas. We discuss several methods to promote road safety and to reduce the incidence and severity of road traffic injuries. Our recommendations take into consideration the incidence and impact of road trauma for different types of road users, and include engineering and infrastructure controls through to interventions targeting or accommodating human behaviour.

Age and engine displacement are associated with unsafe motorcycle rider actions.

BACKGROUND: Our goal was to examine the relationship between age and engine displacement in cubic centimeters (CCs) and crash responsibility. METHODS: Male motorcyclists, aged 16-94, involved in a fatal crash in the United States (1987-2015) who tested negative for both drugs and alcohol were included. Employing a case control design, cases had committed one or more Unsafe Motorcyclist Actions (UMAs), the proxy measure of responsibility; controls had no UMAs recorded. Odds ratios were computed via multinomial regression examining the effect of motorcyclists' age and motorcycle displacement (up to 1500 CCs, in 250 CC increments) on crash responsibility by any UMA and top three individual UMAs committed. RESULTS: A total of 19,166 motorcyclists met our inclusion criteria. Increased displacement was observed in older motorcyclists and in more recent crashes. Fifty-six percent of motorcyclists committed one or more UMAs (n = 10,743). The top three individual UMAs were: Speeding (35%, n = 6,728), Weaving (24%, n = 3,269), and Erratic Operation (6%, n = 1,162). Odds ratios for committing any UMA were the greatest for riders on 750 CC motorcycles, followed closely by 500 and 1000 CC motorcycles. By 1250 CCs the effect of displacement on rider crash responsibility (any UMA) was no longer statistically significant. Typically, younger ages (e.g., 20-30) on motorcycles with 500-1000 CCs were associated with the highest odds of either speeding, weaving, or erratic riding compared to similar aged riders on 250 CC motorcycles. Exceptions were observed, for example riders at 70 years of age on 1500 CCs having higher odds of speeding than younger riders on equivalent CC motorcycles. CONCLUSION: Education and legislative measures should be considered. Educationally, the development of training interventions focusing on control, stability, and breaking differences with more powerful motorcycles (750 to 1250 CCs) is needed. Legislatively, licensing tiers could be employed based on displacement and educational requirements. Education and legislative measures could help to curb the trend seen between high-powered motorcycles and crash responsibility.

Study of work zone traffic safety under adverse driving conditions with a microscopic traffic simulation approach.

Work zone traffic safety under adverse weather conditions has been a serious concern for drivers and transportation agencies. Existing studies on work zone traffic safety with statistical approaches are limited by the availability of data from historical crashes. To date, there is no comprehensive simulation framework to assess work zone traffic safety under adverse driving environments by considering both multi-vehicle and single-vehicle crashes. To fill this gap, this paper presents an integrated framework to evaluate traffic safety in work zone under adverse driving conditions by considering specific work zone configuration, weather and road surface conditions. A new risk index is introduced to assess the traffic safety risk of work zones by integrating the risks of multi-vehicle crashes and single-vehicle crashes. Traffic safety of a typical work zone under different weather conditions is studied to demonstrate the proposed framework. The impacts of the differential speed limits (DSL) and truck proportion on the work zone traffic safety are also investigated. Results show that adverse weather may increase the crash risk in work zones. The effect of DSL on the work zone traffic safety is found to be insignificant while the truck ratio influences the work zone safety in the rainy and snowy weather by primarily affecting the multi-vehicle crash risks.

Traffic volume and crashes and how crash and road characteristics affect their relationship - A meta-analysis.

The present study has investigated the relationship between traffic volume and crash numbers by means of meta-analysis, based on 521 crash prediction models from 118 studies. The weighted pooled volume coefficient for all crashes and all levels of crash severity (excluding fatal crashes) is 0.875. The most important moderator variable is crash type. Pooled volume coefficients are systematically greater for multi vehicle crashes (1.210) than for single vehicle crashes (0.552). Regarding crash severity, the results indicate that volume coefficients are smaller for more fatal crashes (0.777 for all fatal crashes) than for injury crashes but no systematic differences were found between volume coefficients for injury and property-damage-only crashes. At higher levels of volume and on divided roads, volume coefficients tend to be greater than at lower levels of volume and on undivided roads. This is consistent with the finding that freeways on average have greater volume coefficients than other types of road and that two-lane roads are the road type with the smallest average volume coefficients. The results indicate that results from crash prediction models are likely to be more precise when crashes are disaggregated by crash type, crash severity, and road type. Disaggregating models by volume level and distinguishing between divided and undivided roads may also improve the precision of the results. The results indicate further that crash prediction models may be misleading if they are used to predict crash numbers on roads that differ from those that were used for model development with respect to composition of crash types, share of fatal or serious injury crashes, road types, and volume levels.

Numerical investigation of occupant injury risks in car-to-end terminal crashes using dummy-based injury criteria and vehicle-based crash severity metrics.

Guardrails were designed to deter vehicle access to off-road areas and consequently prevent hitting rigid fixed objects alongside the road (e.g. trees, utility poles, traffic barriers, etc.). However, guardrails cause 10 % of deaths in vehicle-to-fixed-object crashes, which recently attracted attention in the highway safety community on the vehicle-based injury criteria used in regulations. The objectives of this study were to investigate both full-body and body-region driver injury probabilities using finite element (FE) simulations, to quantify the influence of pre-impact conditions on injury probabilities, and to analyze the relationship between the vehicle-based crash severity metrics currently used in regulations and the injury probabilities assessed using dummy-based injury criteria. A total of 20 FE impact simulations between a car (Toyota Yaris) with a Hybrid III M50 dummy model in the driver seat and an end terminal model (ET-Plus) were performed in various configurations (e.g. pre-impact velocities, offsets, and angles). The driver's risk of serious injuries (AIS 3+) was estimated based on kinematic and kinetic responses of the dummy during the crashes. A non-linear regression approach was used to compare the injury probabilities assessed in this study to the vehicle-based crash severity metrics used in the testing regulations. In particular, the US Manual for Assessing Safety Hardware (MASH) guideline and European procedures (EN1317) were used for the study. All the recorded dummy-based injury criteria values pass the Federal Motor Vehicle Safety Standard (FMVSS) 208 limits which indicated a low driver risk of serious injury. Overall, the pre-impact vehicle velocity showed to have the highest influence in almost all injury probabilities (59 %, 79 %, 62 %, and 44 % in full-body, head, neck, and chest injuries, respectively). The offset between vehicle midline and the guardrail barrier was the most important variable for thigh injuries (56 %). The assessed injury probabilities were compared to vehicle-based severity metrics. The full-body and chest injuries showed the highest correlation with Occupant Impact Velocity (OIV), Acceleration Severity Index (ASI), and Theoretical Head Impact Velocity (THIV) (R2 > 0.6). Lower correlations of thigh injuries were recorded to OIV (R2 = 0.59) and THIV (R2 = 0.46). Meanwhile, weak correlations were observed between all the other regressions which indicated that no vehicle-based criteria could be used to predict head and neck injuries. Car-to-end terminal crash FE simulations involving a dummy model were performed for the first time in this study. The results pointed out the limitations of the standard vehicle-based injury methods in terms of head and neck injury prediction. The dummy-based injury assessment methodology presented in this study could supplement the crash tests for various impact conditions. In addition, the models could be used to design new advanced guardrail end terminals.

Influences of impact scenarios and vehicle front-end design on head injury risk of motorcyclist.

Motorcycle to vehicle collision is one of the most common accidents in the world and usually leads to serious or fatal head injuries to motorcyclists. This study aims to investigate the influences of impact scenarios and vehicle front-end design parameters on head injury risk of the motorcyclist. Five general vehicle types and different impact scenarios were selected for a parametric analysis. Impact scenarios were set according to ISO, 13232 regulation considering impact angles and impact speeds. Five vehicle types of Sedan, MPV (Multi-Purpose Vehicle), SUV (Sport Utility Vehicle), EV (Electric Vehicle) and 1-Box vehicle were included. HIC15 (Head Injury Criterion), head angular acceleration and CSDM (Cumulative Strain Damage Measure) were calculated to evaluate head injury risk of the motorcyclist. The results show that the critical impact speed for HIC15 and head angular acceleration was around 15 m/s, while the critical speed for CSDM was approximately 10 m/s. Impact angle of 45° show extremely high injury risk to the motorcyclist head. Bonnet leading edge height and its combination with other parameter present high influences on motorcyclist head injuries, and the increasing the bonnet leading edge height can potentially reduce head injury risk of motorcyclists. In summary, the present research results provide some theoretic bases for determining the test speed in motorcycle-vehicle crash regulation and design consideration for typical vehicle front end shape.

Real-time traffic accidents post-impact prediction: Based on crowdsourcing data.

Traffic accident management is a critical issue for advanced intelligent traffic management. The increasingly abundant crowdsourcing data and floating car data provide new support for improving traffic accident management. This paper investigates the methods to predict the complicated behavior of traffic flow evolution after traffic accidents using crowdsourcing data. Based on the available data source, the traffic condition is divided into four levels by congestion delay index: severely congested, congested, slow moving and uncongested. Four types of accidents are consequently defined based on the occurrence of each level. A hierarchical scheme is designed for identifying the most congested level and sequentially predicting duration of each level. The proposed model is validated using traffic accident data in 2017 from an anonymous source in Beijing, China by embedding three machine learning algorithms, random forest (RF), support vector machine (SVM) and neural network (NN), in the scheme. The results show NN outperforms the other two models when the assessment is conducted in absolute differences. Meanwhile, RF has a slightly better performance than SVM, especially when predicting the short-period congestion of severely congested level at the first time. By continuously updating the traffic condition information, significant improvement in accuracy can be acquired regardless of the exact model used. This study shows that emerging crowdsourcing data can be used in a real-time analysis of traffic accidents and the proposed model is effective to analyze such data.

Use of codes data to improve estimates of at-fault risk for elderly drivers.

The fastest-growing demographic in the United States is people aged 65 and over. Because elderly drivers may experience decline in the physical and mental faculties required for driving (which could lead to unsafe driving behaviors), it is critical to determine whether elderly drivers are more likely than younger drivers to be at fault in a crash. This study uses Kentucky crash data and linked hospital and emergency department records to evaluate whether linked data can more accurately estimate the crash propensity of elderly drivers to be at-fault in injury crashes. The Kentucky crash data is edited to conform to the General Use Model (GUM), with crash propensities for linked data compared to propensities developed using the GUM dataset alone. The quasi-induced exposure method is used to determine crash exposure. Factors such as age, gender, and crash location are explored to assess their influence on the risk of a driver being at fault in an injury crash. The overall findings are consistent with previous research - elderly drivers are more likely than younger drivers to be at fault in a crash. Linking crash with hospital and emergency department records could also establish a clearer understanding of the injury crash propensity of all age groups. Equipped with this knowledge, transportation practitioners can design more targeted and effective countermeasures and safety programs to improve the safety of all motorists.

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.

Understanding crashes involving roadway objects with SHRP 2 naturalistic driving study data.

INTRODUCTION: Crashes involving roadway objects and animals can cause severe injuries and property damages and are a major concern for the traveling public, state transportation agencies, and the automotive industry. This project involved an in-depth investigation of such crashes based on the second Strategic Highway Research Program (SHRP 2) Naturalistic Driving Study (NDS) data including detailed information and videos about 2,689 events. METHODS: The research team conducted a variety of logistic regression analyses, complemented by Support Vector Machine (SVM) analyses and detailed case studies. RESULTS: The logistic regression results indicated that driver behavior/errors, involvement of secondary tasks, roadway characteristics, lighting condition, and pavement surface condition are among the factors that contributed significantly to the occurrence and/or increased severity outcomes of crashes involving roadway objects and animals. Among these factors, improper turning movements (odds ratio = 88), avoiding animal or other vehicle (odds ratio = 38), and reaching/moving object in vehicle (odds ratio = 29) particularly increased the odds of crash occurrence. Factors such as open country roadways, sign/signal violation, unfamiliar with roadway, fatigue/drowsiness, and speeding significantly increased the severity outcomes when such crashes occurred. The sensitivity analysis of the three SVM classifiers confirmed that driver behavior/errors, critical speed, struck object type, and reaction time were major factors affecting the occurrence and severity outcomes of events involving roadway objects and animals. Practical Applications: The study provides insights on risk factors influencing safety events involving roadway objects, including their occurrence and the severity outcomes. The findings allow researchers and traffic engineers to better understand the causes of such crashes and therefore develop more effective roadway- and vehicle- based countermeasures.

Simulating the relative influence of tire, vehicle and driver factors on forward collision accident rates.

INTRODUCTION: There is currently a strong focus within the automotive industry centered on traffic safety, with topics such as distracted driving, accident avoidance technologies, and autonomous vehicles. These papers tend to focus on the possible improvements from a single factor. However, there are many factors that are present in each accident, and it is important to understand the influence of each factor on the relative accident risk in order to identify the most effective approaches for improving driver safety. Rear-end accidents tend to be the most common accident type with approximately 1.8 M cases, or 31% of all accidents, in 2012, according to NHTSA. Of the rear-end accident scenarios, approximately 18-23% occur on wet surfaces. METHOD: A Monte Carlo Forward Collision Simulation models the conditions of a wet rear-end accident and estimates the relative impact of various vehicle collision parameters. The model takes distributions of these parameters as inputs, and outputs a risk of collision relative to a known reference case. The parameters that can be studied include: tire grip level, road grip level, vehicle velocity, following distances, and the presence of vehicle technologies (ABS, FCW & AEB). Distributions of some of these parameters have been improved thanks to Naturalistic Driving Study data from SHRP2. RESULTS: This study shows that these vehicle systems have a large impact on safety and can change the amount of influence attributed to other parameters such as tire grip levels. As the use of automated vehicle systems expands, so will the influence of tire grip performance levels on collision risks. Practical Applications: It is more important than ever for consumers and auto manufacturers to consider tire performance levels. Therefore, the tire industry should continue to focus on wet grip as a key performance related to safety and should strive to continue to improve tire performance.

Near crash characteristics among risky drivers using the SHRP2 naturalistic driving study.

PROBLEM: Previous research have focused extensively on crashes, however near crashes provide additional data on driver errors leading to critical events as well as evasive maneuvers employed to avoid crashes. The Strategic Highway Research Program 2 (SHRP2) Naturalistic Driving Study contains extensive data on real world driving and offers a reliable methodology to study near crashes. The current study utilized the SHRP2 database to compare the rate and characteristics associated with near crashes among risky drivers. METHODS: A subset from the SHRP2 database consisting of 4,818 near crashes for teen (16-19 yrs), young adult (20-24 yrs), adult (35-54 yrs), and older (70+ yrs) drivers was used. Near crashes were classified into seven incident types: rear-end, road departure, intersection, head-on, side-swipe, pedestrian/cyclist, and animal. Near crash rates, incident type, secondary tasks, and evasive maneuvers were compared across age groups. For rear-end near crashes, near crash severity, max deceleration, and time-to-collision at braking were compared across age. RESULTS: Near crash rates significantly decreased with increasing age (p < 0.05). Young drivers exhibited greater rear-end (p < 0.05) and road departure (p < 0.05) near crashes compared to adult and older drivers. Intersection near crashes were the most common incident type among older drivers. Evasive maneuver type did not significantly vary across age groups. Near crashes exhibited a longer time-to-collision at braking (p < 0.01) compared to crashes. SUMMARY: These data demonstrate increased total near crash rates among young drivers relative to adult and older drivers. Prevalence of specific near crash types also differed across age groups. Timely execution of evasive maneuvers was a distinguishing factor between crashes or near crashes. Practical Applications: These data can be used to develop more targeted driver training programs and help OEMs optimize ADAS to address the most common errors exhibited by risky drivers.

Using SHRP2 naturalistic driving data to examine driver speeding behavior.

PROBLEM: Speeding-related crashes continue to be a serious problem in the United States. According to the National Highway Traffic Safety Administration, 26% of all fatal crashes in 2017 had speeding as a contributing factor. METHOD: Vehicle speed data recorded during the Strategic Highway Research Program 2 (SHRP 2) Naturalistic Driving Study were analyzed to identify the frequency of speeding episodes. Up to 100 trips were sampled from 2,910 individual drivers aged 16-64. Vehicle speed data from individual trips were parsed into continuous speeding episodes (SEs) and Free-Flow Episodes (FFEs), which approximated opportunities to speed. RESULTS & DISCUSSION: Driving 10 mph above the posted speed limit (PSL) was common, and 99.8% of drivers had at least one occurrence SE within their trip sample, yielding an average of 2.75 SEs per trip (623,202 SEs in total). The analysis focused on a subset of higher-speed SEs in which the vehicle reached speeds of at least 15 mph above the PSL during the SE (71,113 SEs in total). Average maximum speeds for most higher-speed SEs ranged between 12 mph to 15 mph above the PSL, and most also lasted less than 2 min. Most drivers spent less than 5% of the FFE time speeding, and only a small number of drivers spent more than 10% of the time speeding. There was also a clear trend towards the younger group having higher overall percentages of SE time relative to FFE time. Practical Applications: The methods and measures developed in this study provide the foundation for future analyses to determine if there are different types of speeding that vary in terms of risky characteristics, and further, if certain drivers are more likely to engage in riskier speeding behavior. Identifying higher-risk speeders is an important step for developing countermeasures and strategies targeting drivers that are at greatest risk of speed-related crashes.

Combined latent class and partial proportional odds model approach to exploring the heterogeneities in truck-involved severities at cross and T-intersections.

Although the fatal rate of passenger vehicle-involved crashes has decreased in the United States, the fatal rate of truck-involved crashes has increased. This has, in recent years, become a more severe problem than that caused by passenger vehicle-involved crashes. More studies need to be conducted in order to investigate factors that impact the severity of truck-involved crashes within specific scenarios. This study identifies and evaluates the factors that affect the severity of the truck-involved crashes at cross and T-intersections in North Carolina from 2005 to 2017. A latent class clustering for data segmentation is implemented to mitigate unobserved heterogeneity inherent in the crash data. Four partial proportional odds models, which include fixed and unfixed parameters, are developed considering the heterogeneous and ordinal nature inherent in severities. Estimated parameters and marginal effects are further investigated for better interpreting the impacts. Results show heterogeneous explanatory variables and associated coefficients for different classes and severity levels, which indicate the superiority of this combined approach to obtaining more specific factors and accurate coefficients that are estimated in different scenarios. Many factors are found to contribute to the severities, and crossroad scenarios are found to be more severe than T-intersections. The top five driving behaviors at intersections that contribute to the severity include disregarded signs, improper lane use, followed too closely, ignored signals, and failure to yield. These behaviors arouse a necessity to amend the traffic laws and strengthen drivers' education while giving further insights to engineering practitioners and researchers.

An analysis of risk factors associated with road crash severities in Namibia.

Road crash is a leading cause of death and disabilities in Namibia and other developing countries. Based on recent trends, the World Health Organization indicated that progress to realize Sustainable Development Goal (SDG) target 3.6 - which calls for a 50% reduction in the number of road traffic deaths by 2020 - remains far from sufficient. To contribute to efforts in reducing road fatalities in Namibia, this study examined risk factors associated with the severity of crashes recorded in the country. Mixed logit modelling methodology was adopted to address the problem of unobserved heterogeneity in injury severity analysis. Model estimation results reveal that collision with pedestrians, head-on collisions, ran-off road collisions and crashes involving high occupancy passenger vehicles were more likely to result in fatalities and severe injuries. The findings and recommendations of this study are expected to enhance countermeasure implementation to reduce road crashes in Namibia.