Crash risk patterns among older bicyclists: Insights from hybrid XGBoost-Cluster Correspondence Analysis

Older bicyclists face disproportionately high risks of severe injuries and fatalities in road crashes, yet limited research has explored the impact of roadway, environmental, and demographic factors contributing to these risks.

This study analyzes six years (2017-2022) of crash data from the Texas Department of Transportation's (TxDOT) Crash Records Information System (CRIS) to investigate crash patterns among older bicyclists aged 55 and above. In this paper, the Hybrid Modeling Approach was applied to extract the high-risk scenario for a crash with the use of XGBoost, Random Forest, and Cluster Correspondence Analysis (CCA). With this approach, a relationship was found to exist between variables such as categorical specific roadway environments, visibility conditions, safety measures, and demographic factors responsible for the critical contribution to injury severity of older bicyclists in crashes.

This analysis identified six unique clusters of older bicyclist crashes, each identifying different combinations: urban intersection-related crashes, fatal crashes on marked lanes and driveway access points, high-speed rural road crashes, low-speed road crashes with limited enforcement, rural hillcrest crashes under poor visibility, and non-intersection crashes in unconventional low-speed areas lacking traffic control devices. Results show that inadequate lighting and traffic control measures, combined with high-speed limits and complex road environments, greatly contribute to the aggravation of crash severity for older bicyclists. These reflect infrastructure improvements, visibility enhancement, and policy measures to enhance safety and mobility for older bicyclists.