A Transparent Hybrid Deep Learning Framework to Assess German Measles Disease in Skin Images

One of the most common forms of illness in the world is a skin ailment. Although it's rather common, it might be hard to diagnose due to factors including skin color, hair color, and the presence of moles. Because skin malignancies are so common, advances in computer-aided diagnosis will provide dermatologists with a potent new tool for making accurate diagnoses. Visual inspection and clinical screening are the first steps in diagnosing skin illnesses; further testing, such as dermoscopy, biopsy, and histopathology, may be necessary in some circumstances. Fine-grained differences in lesions make automatic categorization of dermatoscopic pictures difficult. This research proposes a hybrid deep learning pre-trained MobileNet model and -based automated system for the identification of measles skin disorders using clinical photos and patient data. The Boosting Machine for Gradients. Using the Kaggle dataset, the suggested technique has achieved a higher rate of accuracy (above 97%) than any of the competing methods. There was a 98.91 percent rate of accuracy, a 97.76 percent rate of recall, an F1 score of 97.25%, and a kappa score of 0.96.