What are the leading causes of fatal and severe injury crashes involving older pedestrian? Evidence from Bayesian network model

Bayesian networks for identifying causal effects of factors on crash injury severity at signalized intersections

Signalized intersections are the areas where traffic crashes with severe injuries frequently happen. Although existing studies have explored the factors affecting crash injury severity at signalized intersections, intricate causal relationships between factors often fail to be captured. Thus, usage of Bayesian network reveals factors contributing to injury severity and the causal relationships between them, with the use of crash data extracted from the Crash Report Sampling System in 2021. The K2 algorithm and Expectation-Maximization algorithms are adopted for structure learning and parameter learning in Bayesian networks, respectively. The results indicate that 1) factors such as speeding, drunk driving, and use of airbags can significantly affect the injury severity, 2) causal relationships exist between distraction, running the red signal, collision type, and crash injury severity, and 3) compared to the random parameter logit model and random forest, Bayesian network has better accuracy in predicting the crash injury severity. The findings can serve to propose effective traffic safety intervention measures to reduce the injury severity of crashes at signalized intersections.

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.

Temporal instability of the determinants of fatal/severe elderly pedestrian injury outcomes in intersections and non-intersections before, during, and after the COVID-19 pandemic

This study examines the variability in the impacts of factors influencing injury severity outcomes of elderly pedestrians (age >64) involved in vehicular crashes at intersections and non-intersections before, during, and after the COVID-19 pandemic. To account for unobserved heterogeneity in the crash data, a random parameters logit model with heterogeneity in the means approach is utilized to analyze vehicle-elderly pedestrian crash data from Seoul, South Korea, occurring between 2018 and 2022. Preliminary transferability tests revealed instability in factor impacts on injury severity outcomes, highlighting the need to estimate individual models across various road segments and time periods. Thus, the dataset was segregated by crash location (intersection/non-intersection) and period (before, during, and after COVID-19), with individual models estimated for each group. Results obtained from the analyses revealed that back injuries positively influenced fatalities at non-intersections after the pandemic and was negatively associated with fatalities at intersections before the pandemic. Additionally, several indicators demonstrated significant instability in their impact magnitudes across different road segments and crash years. During the pandemic, head injuries increased the probability of fatalities higher at non-intersections. After the pandemic, crosswalk locations decreased the possibility of fatalities more at intersections. Compared to intersection segments, the female indicator reduced the likelihood of fatal injuries at non-intersections more before, during, and after the pandemic. Before the pandemic, much older pedestrians experienced a greater decline in fatalities at intersections than non-intersections. This instability could be attributed to altered mobility patterns stemming from the COVID-19 pandemic. Overall, the study findings highlight the variability of determinants of fatal/severe injury outcomes among elderly pedestrians across various road segments and years, with the underlying cause of this fluctuation remaining unclear. Furthermore, the findings revealed that accounting for heterogeneity in the means of random parameters enhances model fit and provides valuable insights for safety professionals. The factor impact variability in the estimated models carries significant implications for elderly pedestrian safety, especially in scenarios where precise projections of the effects of alternative safety measures are essential. Road safety experts can leverage these findings to refine or update current policies to enhance elderly pedestrian safety at intersections and non-intersections.

Aberrant behaviors of drivers involved in crashes and related injury severity: Are there variations between the major cities in the same country?

INTRODUCTION

Crash data analyses based on accident datasets often do not include human-related variables because they can be hard to reconstruct from crash data. However, records of crash circumstances can help for this purpose since crashes can be classified considering aberrant behavior and misconduct of the drivers involved.

METHOD

In this case, urban crash data from the 10 largest Italian cities were used to develop four logistic regression models having the driver-related crash circumstance (aberrant behaviors: inattentive driving, illegal maneuvering, wrong interaction with pedestrian and speeding) as dependent variables and the other crash-related factors as predictors (information about the users and the vehicles involved and about road geometry and conditions). Two other models were built to study the influence of the same factors on the injury severity of the occupants of vehicles for which crash circumstances related to driver aberrant behaviors were observed and of the involved pedestrians. The variability between the 10 different cities was considered through a multilevel approach, which revealed a significant variability only for the inattention-related crash circumstance. In the other models, the variability between cities was not significant, indicating quite homogeneous results within the same country.

RESULTS

The results show several relationships between crash factors (driver, vehicle or road-related) and human-related crash circumstances and severity. Unsignalized intersections were particularly related to the illegal maneuvering crash circumstance, while the night period was clearly related to the speeding-related crash circumstance and to injuries/casualties of vehicle occupants. Cyclists and motorcyclists were shown to suffer more injuries/casualties than car occupants, while the latter were generally those exhibiting more aberrant behaviors. Pedestrian casualties were associated with arterial roads, heavy vehicles, and older pedestrians.

Crash severity analysis: A data-enhanced double layer stacking model using semantic understanding

The crash severity analysis is of significant importance in traffic crash prevention and emergency resource allocation. A range of innovations offers potential traffic crash severity prediction models to improve road safety. However, the semantic information inherent in traffic crash data, which is crucial in enabling a deeper understanding of its underlying factors and impacts, has yet to be fully utilized. Moreover, traffic crash data are commonly characterized by a small sample size, which leads to sample imbalance problem resulting in prediction performance decline. To tackle these problems, we propose a semantic understanding-based data-enhanced double-layer stacking model, named EnLKtreeGBDT, for crash severity prediction. Specifically, to fully leverage the inherent semantic information within traffic crash data and analyze the factors influencing crashes, we design a semantic enhancement module for multi-dimensional feature extraction. This module aims to enhance the understanding of crash semantics and improve prediction accuracy. Then we introduce a data enhancement module that utilizes data denoising and migration techniques to address the challenge of data imbalance, reducing the prediction model's dependence on large sample crash data. Furthermore, we construct a two-layer stacking model that combines multiple linear and nonlinear classifiers. This model is designed to augment the capability of learning linear and nonlinear mixed relationships, thereby improving the accuracy of predicting the severity of crashes on complex urban roads. Experiments on historical datasets of UK road safety crashes validate the effectiveness of the proposed model, and superior performance of prediction precision is achieved compared with the state-of-the-arts. The ablation experiments on both semantic and data enhancement modules further confirm the indispensability of each module in the proposed model.

Identifying the latent relationships between factors associated with traffic crashes through graphical models

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

Do factors associated with older pedestrian crash severity differ? A causal factor analysis based on exposure level of pedestrians

OBJECTIVE

Older pedestrians are more likely to have severe or fatal consequences when involved in traffic crashes. Identifying the factors contributing to the severity and possible interdependencies between factors in specific exposure areas is the first step to improving safety. Therefore, examining the causal factors' impact on pedestrian-vehicle crash severity in a given area is vital for formulating effective measures to reduce the risk of pedestrian fatalities and injuries.

METHODS

This study implements the Thiessen polygon algorithm deployed to define older pedestrians' exposure influence area. Enabling trip characteristics and built environment information as exposure index settings for the background of the pedestrian severity causal analysis. Then, structural equation modeling (SEM) was applied to conduct a factor analysis of the crash severity in high- and low-exposure areas. The SEM evaluates latent factors such as driver risk attitude, risky driving behavior, lack of risk perception among older pedestrians, natural environment, adverse road conditions for driving or walking, and vehicle conditions. The SEM crash model also establishes the relationship between each latent factor.

RESULTS

In total, drivers' risky driving behavior (0.270, p < 0.05) in low-exposure areas significantly impacts older pedestrian crash severity more than in high-exposure areas. Lack of risk perception among older pedestrians (0.232, p < 0.05) is the most critical factor promoting crash severity in high-exposure areas. The natural environment (0.634, p < 0.05) in high-exposure areas positively influences older pedestrians' lack of risk perception more than in low-exposure areas.

CONCLUSIONS

Significant group differences (p-values ∼ 0.001-0.049) existed between the causal factors of the high-exposure risk areas and the low-exposure risk factors. Different exposure intervals require detailed scenarios based on the critical risks identified. The crash severity promotion measures in different exposure areas can be focused on according to the critical causes analyzed. Those clues, in turn, can be used by transportation authorities in prioritizing their plans, policies, and programs toward improving the safety and mobility of older pedestrians.

A Coupled Mathematical Model of the Dissemination Route of Short-Term Fund-Raising Fraud

To effectively protect citizens’ property from the infringement of fund-raising fraud, it is necessary to investigate the dissemination, identification, and causation of fund-raising fraud. In this study, the Susceptible Infected Recovered (SIR) model, Back-Propagation (BP) neural network, Fault tree, and Bayesian network were used to analyze the dissemination, identification, and causation of fund-raising fraud. Firstly, relevant data about fund-raising fraud were collected from residents in the same area via a questionnaire survey. Secondly, the SIR model was used to simulate the dissemination of victims, susceptibles, alerts, and fraud amount; the BP neural network was used to identify the data of financial fraud and change the accuracy of the number analysis of neurons and hidden layers; the fault-tree model and the Bayesian network model were employed to analyze the causation and importance of basic events. Finally, the security measures of fund-raising fraud were simulated by changing the dissemination parameters. The results show that (1) for the spread of the scam, the scale of the victims expands sharply with the increase of the fraud cycle, and the victims of the final fraud cycle account for 12.5% of people in the region; (2) for the source of infection of the scam, the initial recognition rate of fraud by the BP neural network varies from 90.9% to 93.9%; (3) for the victims of the scam, reducing fraud publicity, improving risk awareness, and strengthening fraud supervision can effectively reduce the probability of fraud; and (4) reducing the fraud rate can reduce the number of victims and delay the outbreak time. Improving the alert rate can reduce victims on a large scale. Strengthening supervision can restrict the scale of victims and prolong the duration of fraud.