A data-limited framework for assessing road safety and resilience on rural motorways using social media and network vulnerability scoring

This study presents a novel methodological framework for assessing the vulnerability and resilience of motorway networks in the absence of conventional traffic and incident data. By combining social media accident reports, geometric road data extracted from shapefiles, and binary logistic regression, the research estimates the probability of crash occurrence across Cyprus’s motorways. Furthermore, emergency response accessibility, queue dissipation time and estimated clearance times are used to compute recovery times under both minor and major incident scenarios. A network vulnerability score (NVS) is developed by integrating crash risk and recovery time through grid search optimization, identifying critical motorway segments. The results indicate that the distance from motorway entrance ramps is the only statistically significant geometric factor affecting accident occurrence, with accident odds decreasing by approximately 0.12 % per additional meter from an entrance point. Additionally, the findings show that motorway segments with high vulnerability scores (i.e., NVS = 0.75–1) are predominantly located near urban centers and major entrance ramps, particularly along the Nicosia–Limassol and Nicosia–Larnaca corridors, providing targeted insights for emergency planning and infrastructure resilience improvement.