Diseases like Tuberculosis (TB), Pneumonia, and COVID-19 are critical to global health and require proper detection for their proper treatment. We suggest in this paper a multiclass classification deep learning-based model for automatic diagnosis of these diseases from chest X-ray (CXR) images. The model utilizes a Convolutional Neural Network (CNN) architecture that was trained on open datasets to label CXRs as COVID-19, Pneumonia, Tuberculosis, and No-Findings. We utilized data preprocessing strategies, such as image resizing and normalization, in addition to stratified splitting of data. The suggested model was validated at an accuracy rate of 98.5%, exhibiting robust performance across all the classes, with precision, recall, and F1-score higher than 96%. The Pneumonia class had the best recall (99.8%), and the No-Findings class had balanced results with 99.4% recall and 99.2% F1-score. The results demonstrating the model’s robustness for real-world use in clinical decision support systems. Future development of this project will concentrate on improving the model’s ability to accommodate different types of input data, such as X-rays, CT scans, and other radiological image formats, in order to increase its flexibility and usefulness in various diagnostic applications.