In this article, we propose a noval predictive apprach for prehospital emergency services (PES) demand. The model leverages a negative binomial regression approach, designed to capture the co-evolution of influencing phenomena, including weather conditions, temporal patterns and previous calls volume. To enhance predictive accuracy, we conducted a detailed analysis of these phenomena, examining their impact on call volume variations. The model is validated using real-world data from the Laurentides and Lanaudière regions in Quebec, achieving superior performance compared to existing methods such as Poisson regression and Multi-Layer Perceptron (MLP) neural networks. This work demonstrates that emergency call volumes are highly predictable when meteorological factors are effectively integrated into forecasting models.