Safety sensitivity to roadway characteristics: A comparison across highway classes

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.

Identifying the causal effects of driver distraction on hazardous actions at intersections based on propensity score weighting

Distracted driving is a threat to traffic safety that can result in more traffic crashes. Although previous studies have been conducted to explore the relationship between driver distraction and hazardous driving actions, few studies are available to identify the causation between them. Thus, the study intended to evaluate the causal effects of distraction on hazardous driving actions at intersections based on the crash data extracted from the Crash Report Sampling System (2021-2022). The multinomial logit model was employed to reveal the factors contributing to driver distraction. Then, propensity score weighting was adopted to balance the factor distributions between distraction and non-distraction cases to identify the causal effects on hazardous actions. Results indicated that 1) the propensity of distraction is relevant to factors such as the driver's age, gender, vehicle type, speed limit, area, weather, and light condition, 2) driver distraction can significantly increase the probability of risky actions including speeding, running red lights, failing to obey stop signs, failing to yield, following too closely, and 3) the causal effects show great diversity for different distraction types. The findings serve to understand the influence mechanism of distraction on specific crash risks and develop countermeasures to reduce distraction and hazardous driving actions.

Fatal crashes involving vehicles with driver warning systems: Identifying risk factors using a correlated random parameters ordered logit modeling approach

Recent advancements in vehicular technology are expected to enhance traffic safety by either warning the drivers or by automating the tasks related to driving to reduce the human driver's involvement. The driver warning systems (DWSs) are designed to warn drivers in unsafe situations such as forward collision, lane departure, or when changing lanes with vehicles in blind spot areas. Although these features are designed to enhance safety, recent crash data shows vehicles with these features are still getting involved in crashes, making it necessary to identify the contributing factors. Further, it also requires research to quantify the benefits of vehicles with one or multiple DWS in terms of safety during multivehicle crashes. This study presents a methodological framework to compare factors affecting fatal crashes involving vehicles with no, one and two DWSs. A three-step method is proposed to incorporate unobserved heterogeneity while modeling. Fixed parameter and correlated random parameter order logit models are employed. The results shows that correlated random parameters ordered logit model outperforms traditional fixed parameter ordered logit model. Vehicles equipped with DWSs are safer than vehicles without DWSs during wet or snowy road conditions, when the vehicle skids laterally or longitudinally, and at intersections. Vehicles with one or both DWSs can reduce drink-and-drive and speeding-related crash involvement than vehicles without DWSs. Female and elderly drivers are at a higher risk while driving a vehicle with one or both DWSs compared to driving a vehicle without DWSs, demanding vehicular modifications.

Heterogeneity in crash data: A cross comparison between ordered probability model and its variant using crashes at suburban type arterial

Transportation researchers have long been using the statistical analysis of traffic crash data to create a proactive awareness of traffic safety, make important decisions about the design of vehicles and highways, and develop and implement safe preventive strategies to improve safety. Despite significant progress toward maintaining and analyzing traffic crash data, researchers still encounter several challenges and methodological barriers when conducting statistical analysis. One of these challenges is dealing with the issue of unobserved heterogeneity in crash data. This study uses state-of-the-art methodologies to model the injury severity of traffic crashes that occurred on a specific road segment, namely, a suburban-type road (STR), simultaneously addressing issues related to unobserved heterogeneity in data. Multiple heterogeneity ordered probit models are evaluated against Ohio crash data from the Highway Safety Information System (HSIS). The findings reveal the heterogeneous nature of some variables, such as the nighttime indicator, and demonstrate the distinctive feature of each model to capture the effect of unobserved heterogeneity in analyzing data with such variables. Furthermore, the result helps comprehend the contextual scenarios of crashes at STRs and formulate practical plans to lower the severity of such crashes.

Vehicle-Mounted Pavement Health Monitoring System Based on a Spring-Guide-Assisted Triboelectric Sensor

The present pavement health monitoring is limited by a professional staff patrol. Herein, a vehicle-mounted pavement health monitoring system (VPHMS) based on a spring-guide-assisted triboelectric sensor (S-TES) is proposed, which uses the vibration generated by vehicles passing through uneven pavements to monitor the pavement damage. The VPHMS consists of S-TES, the integration module, and the terminal display module. The designed S-TES has high stability and can achieve a high linear vibration amplitude measurement within 90 mm. Moreover, the integration module is used to process signals and transmit wireless data transmission. The terminal display module is used to receive signals and display the measurement results on the screen. When a vehicle equipped with VPHMS is driven over uneven pavements, the system can accurately monitor the potholes and upheavals on the pavement in real time. This work has significant application value in fields such as pavement health monitoring and intelligent transportation.

Risk exposure factors influencing the frequency of road crashes during the COVID-19 pandemic in Ciudad Juarez, Mexico. A negative binomial spatial regression model

The article aims to investigate the influence of risk exposure factors on the frequency of road crashes from January to August 2020 in Ciudad Juarez, Mexico. It is a longitudinal study with four data sets: road crashes, population and housing census, location of economic activities, and road network information. Specifically, this study investigates the relationship between exposure factors - demographics, main roads and land use - and road crashes. A mixed method analysis was employed, (1) spatial analysis using GIS techniques; and (2) a negative binomial spatial regression model. The results showed a strong spatial dependence (0.274; p-value 0.00) of road crashes in the census tracts, and this effect was statistically significant (0.007) in the spatial regression model. In the model, a high probability (<0.05) of road crashes in the census tracts was found with the population aged 15 to 65 years, the length of main roads and the level of road coverage (Engel index), land uses with economic activities of an industrial and commercial character. The findings of this study successfully capture the social, economic, and urban conditions during the January-August 2020 period in the context of the COVID-19 pandemic. This new knowledge could help create preventive plans and policies to address the frequency of road crashes.

Copula-based bivariate count data regression models for simultaneous estimation of crash counts based on severity and number of vehicles

Statistical models of crash frequency typically apply univariate regression models to estimate total crash frequency or crash counts by various categories. However, a possible correlation between the dependent variables or unobserved variables associated with the dependent variables is not considered when univariate models are used to estimate categorized crash counts-such as different severity levels or numbers of vehicles involved. This may lead to inefficient parameter estimates compared to multivariate models that directly consider these correlations. This paper compares the results obtained from univariate negative binomial regression models of property-damage only (PDO) and fatal plus injury (FI) crash frequencies to models using traditional bivariate and copula-based bivariate negative binomial regression models. A similar comparison was made using models for the expected crash frequency of single- (SV) and multi-vehicle (MV) crashes. The models were estimated using two-lane, two-way rural highway segment-level data from an engineering district in Pennsylvania. The results show that all bivariate negative binomial models (with or without copulas) outperformed the corresponding univariate negative binomial models for PDO and FI, as well as SV and MV, crashes. Second, the statistical association of various traffic and roadway/roadside features with PDO and FI, as well as SV and MV crashes, were not the same relative to their corresponding relationships in the univariate models. The bivariate negative binomial model with normal copula outperformed all other models based on the goodness-of-fit statistics. The results suggest that copula-based bivariate negative binomial regression models may be a valuable alternative for univariate models when simultaneously modeling two disaggregate levels of crash counts.

An investigation of brake failure related crashes and injury severity on mountainous roadways in Wyoming

INTRODUCTION

Although the braking system plays a key role in a safe and smooth vehicular operation, it has not been given proper attention and hence brake failures are still underrepresented in traffic safety. The current body of literature on brake failure-related crashes is very limited. Moreover, no previous study was found to extensively investigate the factors associated with brake failures and the corresponding injury severity. This study aims to fill this knowledge gap by examining brake failure-related crashes and assessing the factors associated with the corresponding occupant injury severity.

METHOD

The study first performed a Chi-square analysis to examine the relationship among brake failure, vehicle age, vehicle type, and grade type. Three hypotheses were formulated to investigate the associations between the variables. Based on the hypotheses, vehicles aged more than 15 years, trucks, and downhill grade segments seemed to be highly associated with brake failure occurrences. The study also applied the Bayesian binary logit model to quantify the significant impacts of brake failures on occupant injury severity and identified various vehicle, occupants, crash, and roadway characteristics.

CONCLUSIONS AND PRACTICAL APPLICATIONS

Based on the findings, several recommendations regarding enhancing statewide vehicle inspection regulation were outlined.

Analysis of driver-injury severity: a comparison between speeding and non-speeding driving crash accounting for temporal and unobserved effects

This paper aims to investigate the differences between temporal stability of factors influencing driver-injury severities in crashes involving speeding and non-speeding driving using a six-year (2012-2017) crash data in Thailand. With two possible driver injury severity outcomes (no/minor and severe/fatal), random parameter binary logit models, that allow for heterogeneity in means and variances, were estimated to fully account for unobserved heterogeneities (i.e., allow crash-level factors to vary across crashes and to influence random parameter distribution). While most factors were unstable over time, speeding crash models result showed that stable factors decreasing probability of severe/fatal injury were restraint, van, passenger car, pickup truck, running-off-road on straight and hitting guardrail and mounting traffic island; whereas stable factor increasing probability of severe/fatal injury were central/eastern/southern regions. In non-speeding driving crash model, stable factors decreasing probability of severe/fatal injury were restraint, truck, and running-off-road on straight and hitting guardrail; whereas stable factors increasing probability of severe and fatal injury were under influence of alcohol and van. The findings of this research could potentially be utilized to improve highway safety and facilitate the development of more effective crash injury mitigation policies. Practical-related recommendation based on the results is also provided.

Why Do Drivers' Collision Avoidance Maneuvers Tend to Cause SUVs to Sideslip or Rollover on Horizontal Curve and Grade Combinations?-An Analysis of the Causes Based on a Modified Multibody Dynamics Model

The extent to which drivers' collision avoidance maneuvers affect the safety margins of sideslip and rollover is not captured by road geometric design theory. To quantify the effects of drivers' collision avoidance maneuvers on the safety margins of sport utility vehicles (SUVs) on horizontal curve and grade combinations, a modified 8-degree-of-freedom multibody model based on SUVs was developed. The model was then used to calculate the design safety margins of sideslip and rollover for steady states and the actual safety margins for collision avoidance maneuvers. Subsequently, the design safety margin reduction rate (the difference between the design and actual safety margins divided by the design safety margin) was calculated and used to assess the safety margins. The results showed that the safety margins of SUVs were significantly reduced by braking, lane changing, and lane changing with braking. The marginal effects indicated that the greater the deceleration and the shorter the lane change duration, the greater the effect on the safety margins, particularly the sideslip safety margin. Furthermore, when the SUV was driven at 80 km·h^-1 on grades with a horizontal curve radius of 270 m and 400 m, the sideslip safety margin with emergency braking (deceleration over -4.5 m·s^-2) was reduced by 71% and 21%, and the rollover safety margin was reduced by 11% and 5%, respectively. Under these conditions, an emergency lane change (lane change duration less than 2 s) caused the SUV to sideslip and reduced the rollover safety margin by 47% (curve radius 270 m) and 45% (curve radius 400 m). Therefore, drivers' collision avoidance maneuvers are a factor that cannot be neglected in alignment design.

Measuring Roadway Lane Widths Using Connected Vehicle Sensor Data

The United States has over three trillion vehicle miles of travel annually on over four million miles of public roadways, which require regular maintenance. To maintain and improve these facilities, agencies often temporarily close lanes, reconfigure lane geometry, or completely close the road depending on the scope of the construction project. Lane widths of less than 11 feet in construction zones can impact highway capacity and crash rates. Crash data can be used to identify locations where the road geometry could be improved. However, this is a manual process that does not scale well. This paper describes findings for using data from onboard sensors in production vehicles for measuring lane widths. Over 200 miles of roadway on US-52, US-41, and I-65 in Indiana were measured using vehicle sensor data and compared with mobile LiDAR point clouds as ground truth and had a root mean square error of approximately 0.24 feet. The novelty of these results is that vehicle sensors can identify when work zones use lane widths substantially narrower than the 11 foot standard at a network level and can be used to aid in the inspection and verification of construction specification conformity. This information would contribute to the construction inspection performed by agencies in a safer, more efficient way.

An empirical analysis of factors contributing to roadway infrastructure damage from expressway accidents: A Bayesian random parameters Tobit approach

This paper presents an empirical analysis of factors contributing to roadway infrastructure damage from expressway accidents, using a Bayesian random parameters Tobit model. The accident data collected from Kaiyang Expressway, China in 2014 and 2015 are used for the empirical analysis. The results of parameter estimation in the proposed model indicate that: the effects of vehicle types are significantly heterogeneous across observations, and that the effects of horizontal curvature, time of day, vehicle registered province, and accident type are also significant but homogeneous across observations. The marginal effects of these contributing factors are calculated to explicitly quantify their impacts on road infrastructure damage. According to the analysis results, some strategies pertaining to safety education, traffic enforcement, roadway design, and intelligence transportation technology are advocated to reduce road infrastructure damage from expressway accidents.

Risk Identification and Conflict Prediction from Videos Based on TTC-ML of a Multi-Lane Weaving Area

Crash risk identification and prediction are expected to play an important role in traffic accident prevention. However, most of the existing studies focus only on highways, not on multi-lane weaving areas. In this paper, a potential collision risk identification and conflict prediction model based on extending Time-to-Collision-Machine Learning (TTC-ML) for multi-lane weaving zone was proposed. The model can accurately learn various features, such as vehicle operation characteristics, risk and conflict distributions, and physical zoning characteristics in the weaving area. Specifically, TTC was used to capture the collision risk severity, and ML extracted vehicle trajectory features. After normalizing and dimensionality reduction of the vehicle trajectory dataset, Naive Bayes, Logistic Regression, and Gradient Boosting Decision Tree (GBDT) models were selected for traffic conflict prediction, and the experiments showed that the GBDT model outperforms two remaining models in terms of prediction accuracy, precision, false-positive rate (FPR) and Area Under Curve (AUC). The research findings of this paper help traffic management departments develop and optimize traffic control schemes, which can be applied to Intelligent Vehicle Infrastructure Cooperative Systems (IVICS) dynamic warning.

An Impact study of highway design on casualty and non-casualty traffic accidents

Background Road Safety has become a worldwide concern due to the alarming repercussions road accidents may bear. This study examined the relationship between different geometric design elements and the accident rates on Rashid Bin Saeed Street, Arabian Gulf Street, and Sultan Bin Zayed Street in Abu Dhabi, United Arab Emirates. Methods The geometric design was collected from the satellite images of google earth in compliance with the standard geometric design manual of Abu Dhabi roads. The recorded geometric data consisted of the number of lanes, lane widths, median length, and width. The traffic volume data was provided by the Integrated Transport Center of Abu Dhabi, which was then converted into Annual Average Daily Traffic (AADT) for analytical purposes. For the studied roads, AADT ranges ranged between 26,509 and 121,890 vehicles per day. The crash data related to the period of 2012-2019 was collected from the online open-access data provided by the United Arab Emirates Ministry of Interior. The data provided had considered variables related to driver gender, age and speed, travel direction, and time of the day amongst other factors. A comprehensive statistical analysis was conducted to study the impact of geometric design elements on road safety through a stable distribution. Stable distributions are generally characterized by four parameters and expressed as X∼S(α,β,σ,μ). The statistical model included several graphical representations such as accident frequency at two levels of severity, casualty and non-casualty accidents for different road segments, traffic volumes, day of the week, age of the injured person, and the geometric design parameters on the three roads. Variance-based methods of sensitivity analysis are also used that are a class of probabilistic approaches that quantify the input and output uncertainties as probability distributions and decompose the output variance into parts attributable to input variables and combinations of variables. The sensitivity of the output to an input variable is therefore measured by the amount of variance in the output caused by that input. Findings The results showed that the accident profiles differ with varying segments on each road, revealing some segments to be of higher accident rates than others. Also, a higher accident frequency was shown with young adult drivers, and a high majority of accidents had occurred on weekends. Regarding the road's geometric design, which is the focus of this study, a sensitivity analysis was made to determine the most influential geometric design element on accident frequency. Interpretation The number of lanes had the highest sensitivity index followed by the median width, and then came the lane width. Thus, modifying the number of lanes on a highway is anticipated to have the highest impact on accident frequency and road safety than any other geometric parameter.

A simulation based large bus side slip and rollover threshold study in slope-curve section under adverse weathers

To study the side slip and rollover threshold of large bus in slope-curve section under adverse weather, factors that affect the safety of large buses that run in slope-curve section, such as rain, snow, cross-wind environmental factors, and road geometry, were analyzed to obtain the friction coefficient of the road surface under different rainfall and snowfall intensities through field measurements and to determine the six-component force coefficient of wind that acts on large buses through wind tunnel tests. The force analysis of large bus in slope-curve section was carried out, and the mechanical equations of large bus under the limit conditions of sideslip and rollover in slope-curve section were established. TruckSim simulation test platform was used to establish a three-dimensional road model and large bus mechanical model at a design speed of 100 km/h. Input parameters, such as cross-wind speed and road friction coefficient, simulate the impact of wind-rain/snow coupling. Under the combined action of wind-rain/snow, the operation test of large bus in slope-curve section was carried out, and the key parameters and indicators of the sideslip and rollover of large bus in slope-curve section were outputted and analyzed. The sudden change point of lateral acceleration is the judging condition for sideslip of large bus in slope-curve section under different road friction coefficient (0.2-0.7), changing from 0.15m/s2 and stabilizing to 0.52 m/s2, and a 0N vertical reaction force of the inner tire is the critical judging condition for rollover under road friction coefficient0.8, and the operating speed thresholds were proposed under different road friction coefficient. This study is expected to provide theoretical support for the speed limit of large bus in slope-curve section under adverse weather.

Analysing the main and interaction effects of commercial vehicle mix and roadway attributes on crash rates using a Bayesian random-parameter Tobit model

In previous research, the effects of commercial vehicle proportions (CVP) on overall crash propensity have been found to be significant, but the results have been varied in terms of the effect direction. In addition, the mediating or moderating effects of roadway attributes on the CVP-vs-safety relationships, have not been investigated. In addressing this gap in the literature, this study integrates databases on crashes, traffic, and inventory for Hong Kong road segments spanning 2014-2017. The classes of commercial vehicles considered are public buses, taxi, and light-, medium- and heavy-goods vehicles. Random-parameter Tobit models were estimated using the crash rates. The results suggest that the CVP of each class show credible effects on the crash rates, for the various crash severity levels. The results also suggest that the interaction between CVP and roadway attributes is credible enough to mediate the effect of CVP on crash rates, and the magnitude and direction of such mediation varies across the vehicle classes, crash severity levels, and roadway attribute type in four ways. First, the increasing effect of taxi proportion on slight-injury crash rate is magnified at road segments with high intersection density. Second, the increasing effect of light-goods vehicle proportion on slight-injury crash rate is magnified at road segments with on-street parking. Third, the association between the medium- and heavy-goods vehicle proportion and killed/severe injury (KSI) crash rate, is moderated by the roadway width (number of traffic lanes). Finally, a higher proportion of medium- and heavy-goods vehicles generally contributes to increased KSI crash rate at road segments with high intersection density. Overall, the findings of this research are expected not only to help guide commercial vehicle enforcement strategy, licensing policy, and lane control measures, but also to review existing urban roadway designs to enhance safety.

A bibliometric analysis of road traffic injury research themes, 1928-2018

Road traffic accidents have become an important social issue worldwide. This study aims to analyse the research status of road traffic injury from 1928 to 2018 and discuss the future research trends. Co-word analysis was applied to analyse 4,184 articles collected from the core collection of Web of Science. Cluster analysis and social network analysis (SNA) were adopted to group keywords, visualize the links between them, and indicate their importance. Strategic diagram was used to reveal the network status of each cluster. The results lead to the following conclusions: (1) 'Road traffic accident', 'injury', 'road safety', 'mortality', and 'risk factor' are at the centre of social network, indicating that these keywords play the most important roles in the field of road traffic injury research. (2) A total of 60 high-frequency keywords are divided into five clusters, namely 'accident causes leading to injury', 'analysis methods', 'health & injury', 'safety management', and 'road traffic', indicating that they are the main sub-fields of road traffic injury research. (3) 'Risk perception' and 'systems theory' are widely discussed topics emerging in recent years. On the basic of the five clusters, valuable references are provided for future research.

Urban road space allocation incorporating the safety and construction cost impacts of lane and footpath widths

INTRODUCTION

Walkability continues to attract great attention from urban planners, designers, and engineers as they recognize not only the merits of pedestrian facilities in terms of the health benefits but also their demerits in terms of accident risk to pedestrians. Wide footpaths improve the pedestrian environment and experience, and thereby motivate travelers to walk as much as possible. However, if footpaths are too wide, they may leave a smaller space for the roadway. On the other hand, wide road lanes may lead to higher road vehicle safety but are costly to construct and maintain and also may leave little space for the footpath. Evidently, for a fixed urban space, what is needed is an optimal balance between the vehicle lane and pedestrian path. This problem is encountered routinely in dense cities including Hong Kong where land availability is severely limited.

METHOD

To address the issue, this paper first establishes safety performance functions (SPFs) for the pedestrian space and the road space, using the random-parameter negative binomial regression. The results indicate the extent to which road lane and footpath width changes are associated with changes in in-vehicle occupant and pedestrian casualties. Then the paper uses the SPFs to develop a methodology for optimizing the width allocations to the road lanes and footpaths, duly considering the user (safety) costs and agency (construction) costs associated with each candidate allocation of the widths. Finally, the paper analyzes the sensitivity of the optimal solution to the relative weights of user cost and agency cost.

RESULTS

When user and agency costs are considered equally important, the optimal lane width is 5.4 m.

CONCLUSION

It is observed that the road space allocation ratio used by the Hong Kong road agency suggests that the agency places a higher weight to user cost compared to agency cost. Practical Application: The findings can help incorporate design-safety relationships, and the stakeholders (agency and users) perspectives in urban road and footpath design.

Safety effects of traffic lane and shoulder widths on two-lane undivided rural roads: A matched case-control study from Norway

This study estimates the effects of lane and shoulder widths on occurrence of head-on and single-vehicle accidents on rural two-lane undivided roads in Norway while considering the differences between winter and non-winter accidents and their severity levels. A matched case-control method was applied to calculate the odds ratios for lane and shoulder width categories, while controlling for the effects of AADT and adjusting for the effects of region, speed limit, segment length, share of long vehicles in AADT and horizontal alignment. The study used a sample of 71,999 roadway segments identified in GIS and 1886 related accidents recorded by the police in five-year period. The results suggest that it is relevant to consider winter and non-winter accidents as well as severe and slight accidents separately when studying the effects of lane and shoulder widths on the occurrence of head-on and single-vehicle accidents. When examining lane and shoulder widths for all related accidents, the lane widths 1.50-2.50 m and shoulder widths 0.50-0.75 m were relatively safer than other categories on Norwegian two-lane rural undivided roads.

The Influence of Road Geometry on Vehicle Rollover and Skidding

This paper analyzes the influence of single and combined unfavorable road geometry on rollover and skidding risks of D-class mid-sized sport utility vehicles (SUVs) with front-wheel drive for roads with design speeds at 80 km/h. A closed-loop simulation model of human-vehicle-road interactions is established to examine the systematic influence of road geometry on vehicle rollover and skidding. The effects of different road geometry on rollover and skidding on SUVs are studied for pavement surface with good and poor friction when vehicles are in the action of steady state cornering. The rollover and skidding risks of the most unfavorable road segments are assessed. The critical wheel is defined by the threshold of skidding during curve negotiation. The results found that SUVs are not easy to rollover on the most unfavorable roads, regardless of good or poor friction of pavement surface. The safety margin of rollover is greater than that of skidding. The safety margin of skidding is minimal on poor friction roads. Therefore, for the sake of driving safety, it is not recommended to design the roads with these unfavorable road geometry combinations.

Risk analysis and management for highway operations safety using a covariate-balanced determinant detector

Highway operations are marred with inherent risks of injury or death, making risk management critical for ensuring the adequate safety of the people involved. This paper investigates the interaction between various highway safety risk factors and effective risk mitigation strategies related to such interaction. The Covariate-Balanced Determinant Detector (CBDD) technique is used to estimate the quantity of both individual and combined risks, and their effect on highway operations safety. Through this technique, the most dangerous risk combinations have been identified and corresponding risk mitigation scenarios have been developed. The results illustrate that the most dangerous scenarios probably result from the interactive effect of risk factors rather than individual factors, and the effect of mitigation strategies should be evaluated in response to a risk scenario before it is implemented.