Injury severity prediction of cyclist crashes using random forests and random parameters logit models.

Cycling provides numerous benefits to individuals and to society but the burden of road traffic injuries and fatalities is disproportionately sustained by cyclists. Without awareness of the contributory factors of cyclist death and injury, the capability to implement context-specific and appropriate measures is severely limited. In this paper, we investigated the effects of the characteristics related to the road, the environment, the vehicle involved, the driver, and the cyclist on severity of crashes involving cyclists analysing 72,363 crashes that occurred in Great Britain in the period 2016-2018. Both a machine learning method, as the Random Forest (RF), and an econometric model, as the Random Parameters Logit Model (RPLM), were implemented. Three different RF algorithms were performed, namely the traditional RF, the Weighted Subspace RF, and the Random Survival Forest. The latter demonstrated superior predictive performances both in terms of F-measure and G-mean. The main result of the Random Survival Forest is the variable importance that provides a ranked list of the predictors associated with the fatal and severe cyclist crashes. For fatal classification, 19 variables showed a normalized importance higher than 5% with the second involved vehicle manoeuvring and the gender of the driver of the second vehicle having the greatest predictive ability. For serious injury classification, 13 variables showed a normalized importance higher than 5% with the bike leaving the carriageway having the greatest normalized importance. Furthermore, each path from the root node to the leaf nodes has been retraced the way back generating 361 if-then rules with fatal crash as consequent and 349 if-then rules with serious injury crash as consequent. The RPLM showed significant unobserved heterogeneity in the data finding four normal distributed indicator variables with random parameters: cyclist age ≥ 75 (fatal prediction), cyclist gender male (fatal and serious prediction), and driver aged 55-64 (serious prediction). The model's McFadden Pseudo R2 is equal to 0.21, indicating a very good fit. Furthermore, to understand the magnitude of the effects and the contribution of each variable to injury severity probabilities the pseudo-elasticity was assessed, gaining valuable insights into the relative importance and influence of the variables. The RF and the RPLM resulted complementary in identifying several roadways, environmental, vehicle, driver, and cyclist-related factors associated with higher crash severity. Based on the identified contributory factors, safety countermeasures useful to develop strategies for making bike a safer and more friendly form of transport were recommended.

Systematic literature review of 10 years of cyclist safety research.

Cyclist safety is a research field that is gaining increasing interest and attention, but still offers questions and challenges open to the scientific community. The aim of this study was to provide an exhaustive review of scientific publications in the cyclist safety field. For this purpose, Bibliometrix-R tool was used to analyse 1066 documents retrieved from Web of Science (WoS) between 2012 and 2021. The study examined published sources and productive scholars by exposing their most influential contributions, presented institutions and countries most contributing to cyclist safety and explored countries open towards international collaborations. A keywords analysis provided the most frequent author keywords in cyclist safety shown in a word cloud with E-bike, behaviour, and crash severity representing the primary keywords. Furthermore, a thematic map of cyclist safety field drafted from the author's keywords was identified. The strategic diagram is divided in four quadrants and, according to both density and centrality, the themes can be classified as follows: 1) motor themes, characterized by high value of both centrality and density; 2) niche themes, defined by high density and low centrality; 3) emerging or declining themes, featured by low value of both centrality and density; and 4) basic themes, distinguished by high centrality and low density. The motor themes (i.e., the main topics in cyclist safety field) crash severity and bike network were further explored. The research findings will be useful to develop strategies for making bike a safer and more confident form of transport as well as to guide researchers towards the future scientific knowledge.

Analysis of contributory factors of fatal pedestrian crashes by mixed logit model and association rules.

Pedestrians are the most vulnerable road users and pedestrian crashes are a major concern both for their number and their severity. In Italy, pedestrians account for 34% of the road fatalities in urban area. To improve pedestrian safety, this study is aimed at analysing the roadway, environmental, vehicle, driver and pedestrian-related factors that are associated with fatal pedestrian crashes in Italy and providing insights for the development of effective countermeasures. This study used an econometric model, the mixed logit model, and a machine learning algorithm, the association rules, to analyse 101,032 pedestrian crashes that occurred in Italy. Study results identified several factors associated with fatal pedestrian crashes. The mixed logit identified 46 significant indicator variables (1 with random parameter), and the association rules provided 119 valid rules. F-measure and G-mean showed higher prediction performance of the mixed logit over the association rules. Study results recommend using both models as complementary approaches since their combination is effective in providing meaningful insights about pedestrian crash contributory factors and their interdependencies. To address the contributory factors identified by the study, behavioural/engineering pedestrian safety countermeasures are recommended. The findings provided new insights for transportation agencies to develop effective countermeasures for pedestrian safety improvement.

Safety Index for evaluation of urban roundabouts.

Recently, there is a growing interest in road safety assessments based on the examination of the characteristics of the road aimed at identifying the presence of risk factors. This approach, named road assessment program or network wide road safety assessment, is required by the EU Directive 2019/1936 on road infrastructure safety management. Reliable procedures for assessing the inherent safety of all the elements of the road network are required to conduct roadway safety assessments. To provide a contribution toward the development of procedures for network wide road safety assessment, this paper develops and validates a Safety Index (SI) for evaluating urban roundabouts. The SI is assessed both at the roundabout level as well as at the roundabout approach level. This procedure detects the safety issues that are the largest contributors to crash risk in order to identify the safety measures that provide the greatest crash reduction at roundabouts. The SI is formulated by combining two components: the exposure of road users to road hazards (Exposure) and the risk factors which increase the probability of involvement in crashes (Risk Index). The procedure considers 33 detailed safety issues and 5 general safety issues to compute the Risk Index. Criteria for identifying and ranking the safety issues are defined. The SI procedure was validated in a sample of 50 urban roundabouts located in Rome, Italy. The sample consisted of 12 single-lane roundabouts and 38 two-lane roundabouts, with a total number of approaches equal to 179. In these roundabouts, the SI scores and the EB crash estimates were compared with reference both to the whole roundabouts as well as to the single roundabout approaches. The correlation between the SI scores and EB estimates was highly significant both at the roundabout level (R2 = 0.85, t = 16.49, p-value < 0.001) as well as at the approach level (R2 = 0.56, t = 14.88, p-value < 0.001). The results from Spearman's rank-correlation analysis provided further validation for the SI indicating that rankings from the SI and the EB estimates agree at the 99.9 % confidence level both at the roundabout level (ρs = 0.80) as well as at the approach level (ρs = 0.70).

Studying the effects of an advanced driver-assistance system to improve safety of cyclists overtaking.

Among all crashes involving cyclists, a motorist approaching from behind a cyclist on a shared lane is particularly dangerous and likely to result in serious injuries and fatalities. Previous research has highlighted that inadequate lateral distance and high vehicle speed are among the main contributing factors of crashes involving cars overtaking cyclists. A new advanced driver assistance system (ADAS) which supports drivers as they overtake cyclists was designed to avoid or, at least, mitigate crashes. In human-machine interface (HMI) design, the information was presented via multiple modalities with a multistage warning system. A combination of lateral clearance (LC) and time-to-danger (TTD) parameters was used as ADAS activation criterion. Experimentation was carried out using the medium-fidelity driving simulator at the Transportation Research Institute (IMOB) of Hasselt University in Belgium. Forty-eight drivers drove the two-lane rural experimental route two times, in baseline condition and with the ADAS activated, testing three overtaking events. Statistical tests showed that the proposed in-vehicle driving assistance system had a significant effect in increasing 1) the length of the passing phase, 2) the LC in the overtaking passing phase, and 3) the TTD along the overtaking maneuver. No effect of the ADAS system on vehicle speed was observed. Overall, the designed system is effective in improving car-cyclist overtaking behaviour in terms of both safety and cyclists' mobility.

Rule discovery to identify patterns contributing to overrepresentation and severity of run-off-the-road crashes.

The main objective of this paper was to analyse the roadway, environmental, and driver-related factors associated with an overrepresentation of frequency and severity of run-off-the-road (ROR) crashes. The data used in this study refer to the 6167 crashes occurred in the section Naples-Candela of A16 motorway, Italy in the period from 2001 to 2011. The analysis was carried out using the rule discovery technique due to its ability of extracting knowledge from large amounts of data previously unknown and indistinguishable by investigating patterns that occur together in a given event. The rules were filtered by support, confidence, lift, and validated by the lift increase criterion. A two-step analysis was carried out. In the first step, rules discovering factors contributing to ROR crashes were identified. In the second step, studying only ROR crashes, rules discovering factors contributing to severe and fatal injury (KSI) crashes were identified. As a result, 94 significant rules for ROR crashes and 129 significant rules for KSI crashes were identified. These rules represent several combinations of geometric design, roadside, barrier performance, crash dynamic, vehicle, environmental and drivers' characteristics associated with an overrepresentation of frequency and severity of ROR crashes. From the methodological point of view, study results show that the a priori algorithm was effective in providing new information which was previously hidden in the data. Finally, several countermeasures to solve or mitigate the safety issues identified in this study were discussed. It is worthwhile to observe that the study showed a combination of factors contributing to the overrepresentation of frequency and severity of ROR crashes. Consequently, the implementation of a combination of countermeasures is recommended.

Crash modification functions for pavement surface condition and geometric design indicators.

The combined contribution to highway safety of pavement surface and geometric design indicators is not well investigated due to the complexity of data collection and high time variability of pavement surface conditions. Introduction of high efficiency equipment for comprehensive road surveys is mitigating this issue, expanding possibilities of data integration. In this framework, the present study developed crash modification functions (CMFs) of pavement surface and geometric design indicators for different crash types (total, run-off-the-road, and others), pavement conditions (dry and wet), and lighting conditions (daytime and nighttime) based on data from two-lane rural highways in Italy. Geometric and pavement data were surveyed with the Automatic Road Analyzer and the Grip Tester. Pavement surface condition data were corrected to the crash time by pavement performance deterioration models based on traffic load to account for the variation in pavement conditions over time. Crash, traffic and weather data were retrieved from national and local databases. This study used safety performance functions (SPFs), fitted with generalized linear modelling techniques and a negative binomial distribution error structure, for developing CMFs. The SPFs were used to quantify the effect of a specific variable on crash occurrence and CMFs were then derived from the model coefficients. CMFs were developed for the following parameters: Grip Number, International Roughness Index, curvature change ratio, coefficient of variation of the curvature, maximum superelevation deficiency, and minimum lane width. According to the study results, an increase in friction, as measured by the Grip Number, is associated with a reduction in crash frequency while an increase in roughness, as measured by the International Roughness Index, is associated with an increase in crash frequency. Thus, both pavement maintenance treatments aimed at increasing friction as well as treatments aimed at reducing irregularities have a positive safety effect, especially when wet, run-off-the-road or nighttime crashes are overrepresented. Study results allow to effectively integrate pavement management systems and safety management systems. When developing paving schedules, transportation agencies often base their decisions on asset management condition targets but do not explicitly account for the role of pavement conditions in highway safety. Availability of CMFs for both pavement surface parameters as well as geometric design parameters is crucial to improve pavement and geometric characteristics considering their safety effects.

A data mining approach to investigate patterns of powered two-wheeler crashes in Spain.

Powered two-wheelers (PTWs) are growing globally each year as they are considered an attractive alternative to cars (flexible, small, affordable, fast and easy to park), especially on congested traffic situations. However, PTWs represent an important challenge for road safety. In fact, in 2016, Spain ranked fifth in terms of PTW fatalities among EU 28. For this reason, this paper aims to investigate which are the patterns among crash characteristics contributing to PTW crashes in Spain. Data from 78,611 crashes involving PTWs occurred in Spain in the period 2011-2013 were analyzed. The analysis was performed by using classification trees and rules discovery which are suitable models aimed at extracting knowledge and identifying valid and understandable patterns from large amounts of data previously unknown and indistinguishable. The response variables assessed in this study were severity and crash type. As a result, several combinations of road, environmental and drivers' characteristics associated with severity and typology of PTW crashes in Spain were identified. Based on the analysis results, several countermeasures to solve or mitigate the safety issues identified in the study were proposed. From the methodological point of view, study results show that both the classification trees and the a priori algorithm were effective in providing non-trivial and unsuspected relations in the data. Classification trees structure allowed a simpler understanding of the phenomenon under study while association discovery provided new information which was previously hidden in the data. Given that the results of the two different techniques were never contradictory, we recommend using classification trees and association discovery as complementary approaches since their combination is effective in exploring data providing meaningful insights about PTW crash characteristics and their interdependencies.

Effects on speed and safety of point-to-point speed enforcement systems: evaluation on the urban motorway A56 Tangenziale di Napoli.

In this paper, we evaluated the effects on speed and safety of the point-to-point (P2P) speed enforcement system activated on the urban motorway A56 in Italy. The P2P speed enforcement is a relatively new approach to traffic law enforcement that involves the calculation of the average speed over a section. To evaluate the speed effects, we performed a before-after analysis of speed data investigating also effects on non-compliance to speed limits. To evaluate the safety effects, we carried out an empirical Bayes observational before-and-after study. The P2P system led to very positive effects on both speed and safety. As far as the effects on the section average travel speeds, the system yielded to a reduction in the mean speed, the 85th percentile speed, the standard deviation of speed, and the proportion of drivers exceeding the speed limits, exceeding the speed limits more than 10km/h, and exceeding the speed limits more than 20km/h. The best results were the decrease of the speed variability and the reduction of the excessive speeding behaviour. The decrease in the standard deviation of speed was 26% while the proportion of light and heavy vehicles exceeding the speed limits more than 20km/h was reduced respectively by 84 and 77%. As far as the safety effects, the P2P system yielded to a 32% reduction in the total crashes, with a lower 95% confidence limit of the estimate equal to 22%. The greatest crash reductions were in rainy weather (57%), on wet pavement (51%), on curves (49%), for single vehicle crashes (44%), and for injury crashes (37%). It is noteworthy that the system produced a statistically significant reduction of 21% in total crashes also in the part of the motorway where it was not activated, thus generating a significant spillover effect. The investigation of the effects of the P2P system on speed and safety over time allowed to develop crash modification functions where the relationship between crash modification factors and speed parameters (mean speed, 85th percentile speed, and standard deviation of speed) was expressed by a power function. Crash modification functions show that the effect of speed on safety is greater on curves and for injury crashes. Even though the study results show excellent outcomes, we must point out that the crash reduction effects decreased over time and speed, speed variability, and non-compliance to speed limits significantly increased over time. To maintain its effectiveness over time, P2P speed enforcement must be actively managed, i.e. constantly monitored and supported by appropriate sanctions.

Analysis of powered two-wheeler crashes in Italy by classification trees and rules discovery.

Aim of the study was the analysis of powered two-wheeler (PTW) crashes in Italy in order to detect interdependence as well as dissimilarities among crash characteristics and provide insights for the development of safety improvement strategies focused on PTWs. At this aim, data mining techniques were used to analyze the data relative to the 254,575 crashes involving PTWs occurred in Italy in the period 2006-2008. Classification trees analysis and rules discovery were performed. Tree-based methods are non-linear and non-parametric data mining tools for supervised classification and regression problems. They do not require a priori probabilistic knowledge about the phenomena under studying and consider conditional interactions among input data. Rules discovery is the identification of sets of items (i.e., crash patterns) that occur together in a given event (i.e., a crash in our study) more often than they would if they were independent of each other. Thus, the method can detect interdependence among crash characteristics. Due to the large number of patterns considered, both methods suffer from an extreme risk of finding patterns that appear due to chance alone. To overcome this problem, in our study we randomly split the sample data in two data sets and used well-established statistical practices to evaluate the statistical significance of the results. Both the classification trees and the rules discovery were effective in providing meaningful insights about PTW crash characteristics and their interdependencies. Even though in several cases different crash characteristics were highlighted, the results of the two the analysis methods were never contradictory. Furthermore, most of the findings of this study were consistent with the results of previous studies which used different analytical techniques, such as probabilistic models of crash injury severity. Basing on the analysis results, engineering countermeasures and policy initiatives to reduce PTW injuries and fatalities were singled out. The simultaneous use of classification trees and association discovery must not, however, be seen as an attempt to supplant other techniques, but as a complementary method which can be integrated into other safety analyses.

Simulator evaluation of drivers' speed, deceleration and lateral position at rural intersections in relation to different perceptual cues.

Aim of the study was to investigate, by means of a driving simulator experiment, drivers' behaviour in terms of speed, deceleration, and lateral position on major approaches of rural intersections in relation to different perceptual cues. In the experiment, ten different design conditions with and without speed-reducing treatments along the approach to the intersection were tested. Twenty-three drivers drove a test route two times and data from the second drive were used for comparison. The order of the ten design conditions was counterbalanced for all the drivers to minimize the presentation order effect. Three different data analysis techniques were used: (a) cluster analysis of speed and lateral position data, (b) statistical tests of speed and lateral position data, and (c) categorical analysis of deceleration behaviour patterns. The most effective treatments were the dragon teeth markings (based on the principle of optical road narrowing), the colored intersection area (based on the principle of intersection highlighting), and the raised median island (based on the principle of physical road narrowing). These measures, in comparison to the base intersection, produced: (1) a significant speed reduction starting from 250 m before the intersection in the range between 13 and 23 km/h, (2) a significant change in the deceleration behaviour with a reduction in the proportion of drivers which did not decelerate, and (3) a shift away from the intersection of the deceleration beginning. Given the significant effects on drivers' behaviour, the dragon teeth markings, the colored intersection area, and the raised median island are strongly recommended for real world implementation.

Traffic calming along rural highways crossing small urban communities: driving simulator experiment.

The paper investigated drivers' speed behaviour in a section of a rural highway crossing a small urban community in the existing scenario without any traffic calming device and in two different design scenarios with traffic calming in the urban community. Two gateways and four integrative traffic calming devices along the route within the urban area were tested. The gateways were aimed at slowing down the vehicles entering in the built-up area, while the traffic calming devices were aimed at complementing the gateway effect inside the built-up area. Two design options were tested: first option (alt1) is a combination of low cost measures, whereas the second option (alt2) is more expensive as includes a chicane and requires land acquisition. Drivers' behaviour was investigated by means of a driving simulator experiment. The VERA dynamic-driving simulator operating at the TEST Road Safety Laboratory located in Naples (Italy) was used. Simulation results were validated by the comparison of speed behaviour in the real world and in the driving simulator, in the scenario without traffic calming. Analysis of the driving simulator experiment results was performed using two different approaches: (a) explorative description of data by cluster analysis; (b) inferential procedures about population using statistical tests. Cluster analysis was carried out in order to test if the drivers' speed behaviour in the different design alternatives was substantially different. Statistical tests were performed in order to verify if speeds in specific sections were significantly different. Cluster analysis looked at speed profiles, whereas statistical tests looked at speed data in specific points. The obtained results showed a different behaviour of drivers approaching the urban community in the existing scenario and in the design scenarios. In the south direction, mean speed reduction ranging between 16 and 17 km/h, with 5% level of significance, was observed. In the north direction, mean speed reduction equal to 11 km/h, with 10% level of significance, was observed. Differences between the two design alternatives were not statistically significant. Along the urban community, a statistically significant mean speed reduction ranging between 9 and 15 km/h was observed in the south direction. In the north direction, speed reduction was not statistically significant. Overall, combined results of cluster analysis and statistical tests showed that the treatments were more effective in the direction with higher speeds in the base scenario.