Early diagnosis of fractures in the bones continues to be a challenge as the new disorders. Rising Artificial Intelligence (AI) algorithms, especially deep learning (DL) methods, have newly turned into a valuable option for detecting fractured bone using X-ray images. Therefore, an advanced deep learning modeling system with X-ray images can identify early diagnosis of various disorders that occur in the bone. We propose an approach which evaluates the efficiency of four pre-trained convolutional neural network (CNN) models, i.e., AlexNet, ResNet-18, GoogLeNet, and SqueezeNet, for the classification of fractured bones from non-fractured bones. Until now, only a FracNet deep learning-based model has been available for rib fracture detection, giving an accuracy of 92.90%. Therefore, we evaluated important factors such as learning rate, type of optimizers, batch size, and no. of epochs to discover the best suitable model. Our analysis of results shows that the ResNet-18 model performs better with a specificity score of 93.98%, an accuracy score of 95.27%, and an area under the ROC curve (AUC) of 0.9382 for fracture classification. In addition, it is observed that the achieved sensitivity, i.e., 99.98%, for bone fracture classification is superior for the ResNet-18 model.