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.

Surrogate measures of safety (SMoS) are meant to be an alternative/complement to crash data that enables a proactive approach to traffic safety. Validity is a fundamental property of SMoS that describes how well they reflect the quality of interest - traffic safety. However, typical validation studies are extremely resource demanding. Therefore, the aim of this study was to explore a relative approach to validity that uses fewer resources and does not rely on crash records from the observed locations. The core idea of a relative approach to the validation of SMoS is that while SMoS might not accurately estimate the expected number of crashes, they still work well for comparisons between different sites or between conditions at the same site. Based on this concept, we propose a method in which a ground truth based on literature can be used as a baseline that can then be compared to the safety analysis of an SMoS study. Using the proposed method, a case study of six intersections in northern Europe was conducted, focusing on bicycle infrastructure. The results indicate that the minimum time-to-collision indicator is in line with the ground truth, whereas the post-encroachment time indicator is not.

Surrogate measures of safety (SMoS) can enable quick, pro-active, and detailed safety evaluations by studying near-crashes. A critical concern regarding SMoS is their validity. This study focused on the validity of two commonly used SMoS indicators-minimum time to collision and post-encroachment time-with a specific focus on bicyclist-motor vehicle interactions. The study was divided into two main parts; the first part focused on observations at intersections in Scandinavia, and the second part focused on developing a crash model using data from 166 similar intersections. Nine signalised intersections in Scandinavia were observed for at least 24 h. During this time, each interaction between a bicyclist and a right- or left-turning motor vehicle was identified and trajectories for the road users were created. The corresponding SMoS values were then calculated. Three main results were found when comparing the results of the crash model with the SMoS analysis. First, there is a significant relationship between the expected number of crashes and both indicators. However, the results also suggest that this correlation might originate from the inherent connection between the indicators and the number of interactions between the studied road users. Finally, when the number of interactions is considered, the results show that the minimum time to collision with a threshold of 3-4 s produces the best results.

Detection and classification of distresses is a fundamental activity in the road pavement management. Even in the early stages of deterioration, road pavement needs to be monitored to identify problems, evaluating the actual conditions and predicting what the future conditions will be. Monitoring activities through manual/visual inspections are time consuming, costly and cause of safety concerns. For these reasons, distress identification is usually limited to few sections randomly selected. The introduction of new high efficiency equipment for distress detection and classification is opening new perspective in road pavement analysis and management. Automatic pavement monitoring and Mechanistic design are introducing new pavement performance indicators and criteria for distress classification. Previous studies show lack of correlations between indexes derived from manual and automatic pavement monitoring. Therefore, capability to derive manual distress parameters from automatic monitoring systems is of great interest in the definition and testing of criteria and methodological approaches. In this paper, a background is reported by referencing examples of North American and Italian tests for the detection and classification of distresses from manual survey and capabilities of the state-of-the-art Automatic Road Analyzer (ARAN 9000) as well. An infield experiment and calibration of a Probabilistic Neural Network Classifier is presented for deriving distress measures from automatic systems.