Among the most dangerous respiratory infections are pneumonia and Chronic Obstructive Pulmonary Disease (COPD). Both those diseases require early, accurate diagnosis in order to receive appropriate treatment. Advances in the deep learning of medical image processing have brought near-clinical reality to the automatic detection of the diseases from chest X-ray images. The paper shall take into consideration the efficiency of CNN-based models in pneumonia and COPD detection using deep learning approaches combined with the incorporation of multimodal data into the system with clinical records for a better diagnosis. We present a performance analysis for variants of the CNN architectures, such as VGG-16, compared in detail regarding their performances on training and testing accuracy, precision, recall, and F1-score. Both pneumonia and COPD provided very promising diagnostic accuracy results, based on such high precision and recall scores showing utility for clinical decision support. This work is illustrative of the great power that multimodal approaches display by introducing imaging and clinical data into further improving diagnosis, especially the differentiation between pneumonia and COPD. By demonstrating an interesting case study of CNNs and multimodal models for accelerating disease detection with a vision toward improving outcome, this paper joins the growing list of publications on applications of artificial intelligence in healthcare. It will also be clear in the future that the results of deep learning methods improve not only in pneumonia detection but also in the viral and obstructive patterns of pulmonary diseases, which would be strong reinforcement for the promise of AI-driven diagnostics in modern health delivery.