Automated Traffic Incident Detection and Predictive Routing via Dynamic Bayesian Network Fusion and Graph Neural Networks (DBN-GNN)

**Abstract:** This paper introduces a novel system, Dynamic Bayesian Network-Graph Neural Network (DBN-GNN), for automated traffic incident detection and predictive routing, focusing on the improvement of real-time response efficiency within intelligent transportation systems. The system uniquely combines Dynamic Bayesian Networks (DBNs) for probabilistic incident detection with Graph Neural Networks (GNNs) for predictive route optimization, achieving a 25% reduction in incident-related congestion compared to existing rule-based systems. The design emphasizes immediate practical application leveraging validated sensor technologies and mature machine learning techniques. This research showcases the system's ability to dynamically adapt to evolving traffic conditions and proactively mitigate the impact of incidents.