Chest X-ray images have been widely and effectively used in the diagnosis of thoracic diseases such as cardiomegaly, pleural effusion, and atelectasis. With the advancement of Artificial Intelligence (AI) technologies, numerous researchers have developed AI-assisted diagnostic systems based on chest X-ray imaging. In this paper, we propose a hybrid machine learning method, termed CFT, designed to diagnose five common chest-related diseases using X-ray images. The core idea behind CFT is to enhance the input data through three augmentation techniques: CutMix, Random Horizontal Flip, and Random Rotation. The model architecture integrates three powerful classification networks—EfficientNet-B3, Feature Pyramid Network (FPN), and a Mini-Transformer—to improve diagnostic performance. Additionally, the model incorporates three types of clinical metadata—age, gender, and view position—during the classification phase to provide contextual enrichment. Beyond image classification, the model is also capable of generating heatmaps to localize lesions. We evaluated CFT on the widely used ChestX-ray14 dataset, with experimental results demonstrating its promising performance. Further discussion and analysis of CFT are presented in the paper.