The early diagnosis of pox symptoms is an important part of preventing a global pandemic. In addition, the estimation of the disease stage and time period of pox rash is of interest. Computer Aided Diagnosis systems have been developed for identification of suspected cases based on machine learning techniques. In recent years, many deep learning approaches have been developed to classify monkeypox disease. In this study, a monkeypox lesion and rash stage classification for self-screening on mobile applications using deep learning techniques was introduced. The datasets consisted of skin lesion and pox rash images. Data augmentation methods were used to increase the sample size and split data into training and testing sets in the experiment setup. When comparing overall accuracy, EfficientNet achieved an accuracy of 0.95 for pox and 0.97 for the pox rash stage. EfficientNet was selected for conversion and implementation on the mobile application. The study determined that the use of deep learning techniques for monkeypox lesions and rash stage classification on a mobile application enabled early identification of patients and effective control of community spread.
