Quantitative decision-support strategies to enhance emergency medical service accessibility in high-risk freeway...

Crashes inside road tunnels often result in more severe outcomes due to spatial constraints that delay prehospital emergency medical services (EMS) access. Emergency evacuation facilities, such as cross passages and emergency parking areas, are critical for ensuring timely EMS response during traffic congestion caused by in-tunnel crashes. This study presents a quantitative approach to develop targeted strategies for enhancing such facilities to improve EMS accessibility in tunnels. Nationwide in-tunnel crash data from Korean freeways over a four-year period were analyzed, and the limited number of severe crash cases was addressed using SMOTE-based data augmentation to balance severity levels in the prediction model. A SHAP-based XGBoost classifier was applied to identify the influence of key features related to EMS accessibility and emergency evacuation facilities on crash severity. The results showed that crashes occurring more than 1 km from the tunnel entrance and EMS response coverage within a 10-15 km range have a significant impact on increasing crash severity. The presence of emergency cross passages, three or more lanes, emergency parking areas, and properly spaced CCTV installations were found to help reduce crash severity. Based on these findings, tunnels with particularly limited EMS accessibility were prioritized, and tailored facility improvement measures were proposed. This study provides quantitative evidence to guide cost-effective upgrades of tunnel emergency evacuation facilities, enhancing prehospital EMS accessibility and emergency preparedness. The proposed methodology can also support decision-making on implementing road tunnel safety improvement policies.