Breast cancer remains a significant global health challenge, especially among women, underscoring the urgent need for advanced diagnostic and prognostic methods. This study explores the capabilities of deep learning (DL) models in classifying mammographic images to aid in the prognosis of breast cancer. Focusing on ResNet50, AlexNet, and a novel hybrid deep learning model, we leveraged the Digital Database for Screening Mammography (DDSM) and its refined variant for model development and evaluation. Our goal was to accurately categorize mammographic images into normal, benign, and malignant classes. Our findings reveal that all examined deep learning architectures exhibited impressive performance on the test set. The ResNet50 model demonstrated a high validation accuracy of 96.23%, while the AlexNet achieved 95.99%. Notably, our hybrid deep learning model outperformed these with an accuracy of 97.23%, showcasing its potential in enhancing the accuracy of breast cancer prognosis. These results suggest that deep learning networks, particularly advanced models like our hybrid model, are effective in identifying mammographic images, which could significantly improve the accuracy of breast cancer prognosis. However, these findings also highlight the necessity for ongoing research. Future studies should aim to further refine these models, possibly through the utilization of larger and more varied datasets, and explore their applicability in clinical environments.