Urban traffic congestion poses significant challenges for today's cities, affecting mobility, productivity, and environmental quality. The present study proposes a data-driven framework that integrates deep learning specifically Recurrent Neural Networks (RNNs) with Digital Twin (DT) technology to enhance travel time prediction and traffic management. The model utilizes real-time and historical data from sources such as Google Maps, weather services, and traffic sensors to capture temporal dynamics and external factors influencing traffic patterns. The RNN model exhibited a high degree of predictive accuracy, as evidenced by its R² value of approximately 0.94. Furthermore, its incorporation into a DT environment facilitated dynamic 3D simulations and route optimization. A comparative analysis revealed that the DT system exhibited a marked superiority over conventional navigation tools in congested scenarios, with a travel time reduction of up to 26%. The findings indicate the potential for a synergistic integration of artificial intelligence (AI) and data technology (DT) to facilitate the development of intelligent, adaptable urban transportation systems.