Predicting driver injury severity and identifying factors influencing crash outcomes are crucial for developing effective traffic safety measures. This study focuses on estimating driver injury severity (uninjured, injured, or killed) and determining critical factors affecting crash outcomes. A hybrid framework combining Deep Neural Networks (DNNs) and Random Forest (RF) is proposed, where a DNN extracts features and RF performs the final classification, leveraging ensemble methods. The results were compared with those of well-known methods (e.g., kNN, XGBoost), with the hybrid approach achieving the best performance (0.92 accuracy, 0.89 F1-macro, 0.91 F1-micro scores) in predicting injury severity. The results showed that crash type, vehicle type, driver fault, intersection type, season, time, and road type had the greatest impact, while factors like pavement condition and driver gender had minimal influence. To the best of our knowledge, this is the first study to combine DNN-based feature extraction with RF classification in the context of traffic injury severity prediction. The framework offers a new approach for drivers and policymakers, providing a deeper understanding of driver injury severity prediction and its underlying factors.

Segregation is one of the quality standards that must be monitored during the fabrication and placement of Portland cement concrete. Segregation refers to separation of coarse aggregate from the cement paste, resulting in inhomogeneous mixture. This study introduces a digital imaging based technique to quantify the segregation of Portland cement concrete from 2D digital images of cut sections. In the previous studies, segregation was evaluated based on the existence of coarse aggregate fraction at different geometrical regions of a sample cross section without considering its distribution characteristics. However, it is shown that almost all particle fractions can form clusters and increase the degree of segregation, thus deteriorating the structural performance of concrete. In the proposed methodology, a segregation index is developed by based on the spatial distribution of different size fractions of coarse aggregate within a sample cross section. It is shown that degradation in mixture’s homogeneity is controlled by the combined effect of particle distribution and their relative proportions in the mixture. Hence, a segregation index characterizing the mixture inhomogeneity is developed by considering not only spatial distribution of aggregate particles, but also their size fractions in the mixture. The proposed methodology can be successfully used as a quality control tool for monitoring the segregation level in hardened concrete samples.

There have been limited efforts to study the severity of traffic crashes in Turkey. This study is probably the first attempt to investigate the factors that contribute to driver injury severity patterns. Fatal injury crash data from 2015 to 2017 for the city of Mersin was used. A two-step approach was employed. First, latent class clustering was performed to capture unobserved heterogeneity inherent in the crash data. The crash database was separated into four clusters by maximizing the homogeneity within each cluster. Then, an ordered probit model was developed on each cluster to explore the factors that significantly affect the driver injury outcomes. Marginal effects were calculated to interpret the influence of significant variables across the injury levels in more detail. The presence of motorcycles, fixed objects, and run-off-road crashes were found to be the main factors associated with injury and fatality in all clusters. The results underlined the association between driving behavior and injury severity of drivers. Alcohol-impaired driving, speeding, and traffic sign/signal violations increase the likelihood of severe injury.