Traditional vehicle routing algorithms do not consider the changing nature of traffic. While implementations of Dijkstra's algorithm with varying weights exist, the weights are often changed after the outcome of algorithm is executed, which may not always result in the optimal route being chosen. Hence, this paper proposes a novel vehicle routing algorithm that improves upon Dijkstra's algorithm using a traffic prediction model based on the traffic flow in a road network. Here, Dijkstra's algorithm is adapted to be dynamic and time dependent using traffic flow theory principles during the planning stage itself. The model provides predicted traffic parameters and travel time across each edge of the road network at every time instant, leading to better routing results. The dynamic algorithm proposed here predicts changes in traffic conditions at each time step of planning to give the optimal forward-looking path. The proposed algorithm is verified by comparing it with conventional Dijkstra's algorithm on a graph with randomly simulated traffic, and is shown to predict the optimal route better with continuously changing traffic.