As urban traffic demand continues to rise, efficient traffic analysis has become increasingly critical. However, traditional methods are often limited to small geographic areas and local analysis. Centrality-based methods are more suitable for global network analysis, but commonly used Betweenness Centrality (BC) and its variants face limitations in road network analysis. Therefore, designing a new BC variant that better reflects traffic flow and supports regional dynamic analysis in large road networks has become essential. To address this, we propose a new BC variant, Region-based OD-Betweenness Centrality (R-ODBC), to integrate actual origin-destination (OD) data at the regional level, effectively supporting the aforementioned needs. Specifically, we first developed an OD-based Betweenness Centrality (ODBC) and designed a precise calculation algorithm. Through theoretical analysis and visualization, we demonstrated a significant correlation between ODBC and actual traffic flow. We further extended this metric to the geographic region, introducing R-ODBC to support more effective regional dynamic analysis. Furthermore, to tackle the high computational complexity of R-ODBC in large road networks, we transformed it into a ranking problem and introduced a Spatio-Temporal Ranking Model (STRM). This model integrates OD interaction, spatial modeling, and historical temporal data, optimized by a spatio-temporal attention mechanism, achieving high precision and efficiency. Experimental results show that our method not only maintains high accuracy but also significantly improves computational efficiency, providing new perspectives for road networks analysis.