This study modeled the crash severity of elderly drivers using data from the state of Virginia, United States, for the period of 2014 through to 2021. The impact of several exogenous variables on the level of crash severity was investigated. A multilevel ordinal logistic regression model (M-OLR) was utilized to account for the spatial heterogeneity across different physical jurisdictions. The findings discussed herein indicate that the M-OLR can handle the spatial heterogeneity and lead to a better fit in comparison to a standard ordinal logistic regression model (OLR), as the likelihood-ratio statistics comparing the OLR and M-OLR models were found to be statistically significant, with p-value of <0.001. The results showed that crashes occurring on two-way roads are likely to be more severe than those on one-way roads. Moreover, the risks for older, distracted, and/or drowsy drivers to be involved in more severe crashes escalate than undistracted and nondrowsy drivers. The data also confirmed that the consequences of crashes involving unbelted drivers are prone to be more severe than those for belted drivers and their passengers. Furthermore, the crash severity on higher-speed roads or when linked to high-speed violations is more extreme than on low-speed roads or when operating in compliance with stated speed limits. Crashes that involve animals are likely to lead to property damage only, rather than result in severe injuries. These findings provide insights into the contributing factors for crash severity among older drivers in Virginia and support better designs of Virginia road networks.

As the global elderly population continues to rise, the risk of severe crashes among elderly drivers has become a pressing concern. This study presents a comprehensive examination of crash severity among this demographic, employing machine learning models and data gathered from Virginia, United States of America, between 2014 and 2021. The analysis integrates parametric models, namely logistic regression and linear discriminant analysis (LDA), as well as non-parametric models like random forest (RF) and extreme gradient boosting (XGBoost). Central to this study is the application of resampling techniques, specifically, random over-sampling examples (ROSE) and the synthetic minority over-sampling technique (SMOTE), to address the dataset’s inherent imbalance and enhance the models’ predictive performance. Our findings reveal that the inclusion of these resampling techniques significantly improves the predictive power of parametric models, notably increasing the true positive rate for severe crash prediction from 6% to 60% and boosting the geometric mean from 25% to 69% in logistic regression. Likewise, employing SMOTE resulted in a notable improvement in the non-parametric models’ performance, leading to a true positive rate increase from 8% to 36% in XGBoost. Moreover, the study established the superiority of parametric models over non-parametric counterparts when balanced resampling techniques are utilized. Beyond predictive modeling, the study delves into the effects of various contributing factors on crash severity, enhancing the understanding of how these factors influence elderly road safety. Ultimately, these findings underscore the immense potential of machine learning models in analyzing complex crash data, pinpointing factors that heighten crash severity, and informing targeted interventions to mitigate the risks of elderly driving.