Analysis of pedestrian-vehicle crash injury severity factors in Colorado 2006-2016

Endogeneity of pedestrian survival time and emergency medical service response time: Variations across disadvantaged and non-disadvantaged communities

The Vision Zero-Safe Systems Approach prioritizes fast access to Emergency Medical Services (EMS) to improve the survivability of road users in transportation crashes, especially concerning the recent increase in pedestrian-involved crashes. Pedestrian crashes resulting in immediate or early death are considerably more severe than those taking longer. The time gap between injury and fatality is known as survival time, and it heavily relies on EMS response time. The characteristics of the crash location may be associated with EMS response and survival time. A US Department of Transportation initiative identifies communities often facing challenges. Six disadvantaged community (DAC) indicators, including economy, environment, equity, health, resilience, and transportation access, enable an analysis of how survival and EMS response times vary across DACs and non-DACs. To this end, this study created a unique and comprehensive database by linking DACs data with 2017-2021 pedestrian-involved fatal crashes. This study utilizes two-stage residual inclusion models with segmentation for DACs and non-DACs accounting for the endogenous relationship between EMS response and pedestrian survival time. The results indicate that EMS response time is higher and pedestrian survival time is lower in DACs than in non-DACs. A delayed EMS response time is associated with a greater reduction in survival time in DACs compared to non-DACs. Factors, e.g., nighttime and interstate crashes, contribute to higher EMS response time, while pedestrian drugs, driver speeding, and hit-and-run behaviors are associated with a greater reduction in survival time in DACs than non-DACs. The implications of the findings are discussed in the paper.

Inherent uncertainty in pedestrian collision reconstruction: How evidence variability affects head kinematics and injury prediction

Reconstructing individual cases from real-world collision data is used as a tool to better understand injury biomechanics and determine injury thresholds. However, real-world data tends to have inherent uncertainty within parameters, such as ranges of impact speed, pre-impact pedestrian stance or pedestrian anthropometric characteristics. The implications of this input parameter uncertainty on the conclusions made from case reconstruction about injury biomechanics and risk is not well investigated, with a 'best-fit' approach more frequently adopted, leaving uncertainty unexplored. This study explores the implications of uncertain parameters in real-world data on the biomechanical kinematic metrics related to head injury risk in reconstructed real-world pedestrian-car collisions. We selected six pedestrian-car cases involving seven pedestrians from the highly detailed GB Road Accident In-Depth Studies (RAIDS) database. The collisions were reconstructed from the images, damage measurements and dynamics available in RAIDS. For each case, we varied input parameters within uncertain ranges and report the range of head kinematic metrics from each case. This includes variations of reconstructed collision scenarios that fits within the constraints of the available evidence. We used a combination of multibody and finite element modelling in Madymo to test whether the effect of input data uncertainty is the same on the initial head-vehicle and latter head-ground impact phase. Finally, we assessed whether the predicted range of head kinematics correctly predicted the injuries sustained by the pedestrian. Varying the inputs resulted in a range of output head kinematic parameters. Real-world evidence such as CCTV footage enabled predicted simulated values to be further constrained, by ruling out unrealistic scenarios which do not fit the available evidence. We found that input data uncertainty had different implications for the initial head-vehicle and latter head-ground impact phase. There was a narrower distribution of kinematics associated with the head-vehicle impact (initial 400 ms of the collision) than in the latter head-ground impact. The mean head-vehicle kinematics were able to correctly predict the presence or absence of both subdural haematoma (using peak rotational acceleration) and skull vault fracture (using peak contact force) in all pedestrians presented. This study helps increase our understanding of the effects of uncertain parameters on head kinematics in pedestrian-car collision reconstructions. Extending this work to a broad range of pedestrian-vehicle collision reconstructions spanning broad population demographics will improve our understanding of injury mechanisms and risk, leading to more robust design of injury prevention measures.

Predicting pedestrian-involved crash severity using inception-v3 deep learning model

This research leverages a novel deep learning model, Inception-v3, to predict pedestrian crash severity using data collected over five years (2016-2021) from Louisiana. The final dataset incorporates forty different variables related to pedestrian attributes, environmental conditions, and vehicular specifics. Crash severity was classified into three categories: fatal, injury, and no injury. The Boruta algorithm was applied to determine the importance of variables and investigate contributing factors to pedestrian crash severity, revealing several associated aspects, including pedestrian gender, pedestrian and driver impairment, posted speed limits, alcohol involvement, pedestrian age, visibility obstruction, roadway lighting conditions, and both pedestrian and driver conditions, including distraction and inattentiveness. To address data imbalance, the study employed Random Under Sampling (RUS) and the Synthetic Minority Oversampling Technique (SMOTE). The DeepInsight technique transformed numeric data into images. Subsequently, five crash severity prediction models were developed with Inception-v3, considering various scenarios, including original, under-sampled, over-sampled, a combination of under and over-sampled data, and the top twenty-five important variables. Results indicated that the model applying both over and under sampling outperforms models based on other data balancing techniques in terms of several performance metrics, including accuracy, sensitivity, precision, specificity, false negative ratio (FNR), false positive ratio (FPR), and F1-score. This model achieved prediction accuracies of 93.5%, 77.5%, and 85.9% for fatal, injury, and no injury categories, respectively. Additionally, comparative analysis based on several performance metrics and McNemar's tests demonstrated that the predictive performance of the Inception-v3 deep learning model is statistically superior compared to traditional machine learning and statistical models. The insights from this research can be effectively harnessed by safety professionals, emergency service providers, traffic management centers, and vehicle manufacturers to enhance their safety measures and applications.

Causal Inference with Case-Only Studies in Injury Epidemiology Research

Purpose of Review

We review the application and limitations of two implementations of the "case-only design" in injury epidemiology with example analyses of Fatality Analysis Reporting System data.

Recent Findings

The term "case-only design" covers a variety of epidemiologic designs; here, two implementations of the design are reviewed: (1) studies to uncover etiological heterogeneity and (2) studies to measure exposure effect modification. These two designs produce results that require different interpretations and rely upon different assumptions. The key assumption of case-only designs for exposure effect modification, the more commonly used of the two designs, does not commonly hold for injuries and so results from studies using this design cannot be interpreted. Case-only designs to identify etiological heterogeneity in injury risk are interpretable but only when the case-series is conceptualized as arising from an underlying cohort.

Summary

The results of studies using case-only designs are commonly misinterpreted in the injury literature.

Understanding the role of faster emergency medical service response in the survival time of pedestrians

Fatalities and severe injuries among vulnerable road users, particularly pedestrians, are rising. In addition to the loss of life, about 6,000 annual pedestrian deaths in the U.S. cost society about $6 billion. Contrary to the assumption that all fatal pedestrian-involved crashes are similar, instantaneous death is substantially more severe than death that occurs several days after the crash. Instead of homogenizing all fatal pedestrian crashes, this study takes into account the severity of fatal injury crashes as a timeline based on the survival time of pedestrians. This study extracts valuable information from fatal crashes by examining pedestrians' survival time ranging from early death to death within 30 days of the crash. The Fatality Analysis Reporting System dataset is utilized from 2015 to 2018. The emergency medical service (EMS) response time is the key post-crash measure, while controlling for pedestrian, driver, roadway, and environmental characteristics. Notably, the response time and survival time can cause endogeneity, i.e., the response times may be shorter for more severe crashes. Due to the spatial and temporal nature of traffic crashes, to extract the association of different variables with pedestrians' survival time, a geographically and temporally weighted truncated regression with a two-stage residual inclusion treatment (local model) is estimated. The local model can overcome the endogeneity limitation (between EMS response time and survival time) and uncover the potentially spatially and temporally varying correlates of pedestrians' survival time with associated factors to account for unobserved heterogeneity. Moreover, to verify the variations are noticeable, a truncated regression with the two-stage residual inclusion treatment is developed (global model). The modeling results indicate that while capturing the unobserved heterogeneity, the local model outperformed the global model. The empirical results show that EMS response time, speeding, and some pedestrian behaviors are the most important factors that affect pedestrians' survival time in fatal injury crashes. However, the effect of factors on pedestrians' survival time is noticeably varied spatially and temporally. The results and their implications are discussed in detail in the paper.

Macro-level literature analysis on pedestrian safety: Bibliometric overview, conceptual frames, and trends

Due to the high volume of documents in the pedestrian safety field, the current study conducts a systematic bibliometric analysis on the researches published before October 3, 2021, based on the science-mapping approach. Science mapping enables us to present a broad picture and comprehensive review of a significant number of documents using co-citation, bibliographic coupling, collaboration, and co-word analysis. To this end, a dataset of 6311 pedestrian safety papers was collected from the Web of Science Core Collection database. First, a descriptive analysis was carried out, covering whole yearly publications, most-cited papers, and most-productive authors, as well as sources, affiliations, and countries. In the next steps, science mapping was implemented to clarify the social, intellectual, and conceptual structures of pedestrian-safety research using the VOSviewer and Bibliometrix R-package tools. Remarkably, based on intellectual structure, pedestrian safety demonstrated an association with seven research areas: "Pedestrian crash frequency models", "Pedestrian injury severity crash models", "Traffic engineering measures in pedestrians' safety", "Global reports around pedestrian accident epidemiology", "Effect of age and gender on pedestrians' behavior", "Distraction of pedestrians", and "Pedestrian crowd dynamics and evacuation". Moreover, according to conceptual structure, five major research fronts were found to be relevant, namely "Collision avoidance and intelligent transportation systems (ITS)", "Epidemiological studies of pedestrian injury and prevention", "Pedestrian road crossing and behavioral factors", "Pedestrian flow simulation", and "Walkable environment and pedestrian safety". Finally, "autonomous vehicle", "pedestrian detection", and "collision avoidance" themes were identified as having the greatest centrality and development degrees in recent years.

Driving behavior at midblock crosswalks with Rectangular Rapid Flashing Beacons: Hidden Markov model approach using naturalistic data

Pedestrian fatalities have increased in the U.S. with the largest increase being observed on urban arterials and away from intersections. Rectangular Rapid Flashing Beacon (RRFB) has been widely implemented as a safety countermeasure to improve driver awareness and visibility of pedestrians, especially for midblock crosswalks. Studies show that drivers are more likely to yield to pedestrians at crosswalks with an RRFB. These studies are often based on a binary outcome of whether or not drivers yield to pedestrians. Nevertheless, it is also important to consider the drivers' deceleration behavior as a dynamic process at these crosswalks and the impact of pedestrians being present or not. Understanding this dynamic behavior and the related circumstances can provide information on the design of alerting systems that help drivers make more appropriate decisions at these crosswalks to avoid a vehicle-pedestrian crash. This study examined this research topic using Hidden Markov Models (HMMs) and data from a naturalistic study. More specifically, four HMMs were applied to the naturalistic brake and jerk data from the Safety Pilot Model Deployment (SPMD) program given drivers' intention to slow down, the RRFB activation status, and the presence of pedestrians. The time-based data sequence was converted to distance-based through a moving window to enhance result comparison and interpretation. Grid-search was used to select the best moving window parameters and the optimal number of hidden states. This study confirmed the high compliance at an activated RRFB when pedestrians were present. Even without pedestrians, one in five traversals showed drivers slowing down to less than 8.94 m/s (20 mph) within 35 m of the crosswalk. Model results further indicate that drivers started braking as far back as 180 m before the crosswalk and stopped braking from 70 m before the crosswalk at an activated RRFB without pedestrians. When there were pedestrians, drivers would start braking 20 to 30 m later but would brake more firmly and for longer. Finally, drivers were not likely to brake or decelerate when RRFB was off and no pedestrians were present.

Safety effects of law enforcement cameras at non-signalized crosswalks: A case study in China

Pedestrians are vulnerable when crossing the street, especially at non-signalized crosswalks. In China, in spite of the priority that laws entitle the pedestrians, the yielding rates at non-signalized crosswalks are relatively low. In light of this situation, law enforcement cameras have been used to increase the percentage of drivers yielding to pedestrians. This study investigates the effectiveness of law enforcement cameras on drivers yielding behavior and vehicle-pedestrian conflicts at non-signalized crosswalks. Using Unmanned Aerial Vehicle (UAV) and roadside video recording, information including pedestrian characteristics, vehicular characteristics and environmental factors are collected. The conflict indicators used include Post-Encroachment Time (PET), Time to Collision (TTC), and Deceleration to Safety Time (DST). In this study, a conflict classification framework based on PET, TTC and DST using Support Vector Machine algorithm is employed. A multinomial logit regression model is used to identify the factors contributing to the conflicts. Then, binary logit regression models are constructed to analyze the effects of law enforcement cameras on drivers yielding behavior. Conflict study reveals that the implementation of law enforcement cameras would increase the probability of slight conflict but decrease the probability of serious conflict. Yielding behavior analysis shows that the illegitimate yielding behavior percentages are over 10 %, indicating the necessity of improving the awareness of yielding rules, and the implementation of law enforcement cameras would increase the yielding and legitimate yielding probability. Moreover, factors including the adjacent vehicle yielding behavior, number of lanes between pedestrian and vehicle, pedestrian speed change, pedestrian waiting time, pedestrian accepted gap time, vehicle upstream speed and vehicle speed change are significantly associated with conflict severity and drivers yielding behavior. We recommend that supplementary facilities and measures should be used to improve the safety performance of law enforcement cameras.

Older Pedestrian Traffic Crashes Severity Analysis Based on an Emerging Machine Learning XGBoost

Older pedestrians are vulnerable on the streets and at significant risk of injury or death when involved in crashes. Pedestrians’ safety is critical for roadway agencies to consider and improve, especially older pedestrians aged greater than 65 years old. To better protect the older pedestrian group, the factors that contribute to the older crashes need to be analyzed deeply. Traditional modeling approaches such as Logistic models for data analysis may lead to modeling distortions due to the independence assumptions. In this study, Extreme Gradient Boosting (XGBoost), is used to model the classification problem of three different levels of severity of older pedestrian traffic crashes from crash data in Colorado, US. Further, Shapley Additive explanations (SHAP) are implemented to interpret the XGBoost model result and analyze each feature’s importance related to the levels of older pedestrian crashes. The interpretation results show that the driver characteristic, older pedestrian characteristics, and vehicle movement are the most important factors influencing the probability of the three different severity levels. Those results investigate each severity level’s correlation factors, which can inform the department of traffic management and the department of road infrastructure to protect older pedestrians by controlling or managing some of those significant features.