The subject matter of the article Prediction of COVID-19 disease and infection rate based on a pre-trained model that supports deep learning. The goal is to build a system to diagnose people as infected or not with covid disease with the percentage of infection and the affected site and to present it with interactive interfaces to facilitate the use of the system for anyone not specialized in the software field. The task is to detect or predict the Coronavirus that affects the airways, lungs, and breathing. It is the cause of many deaths and is still in the process of transformation and development, but with less media exposure. From this standpoint, a medical system was proposed to detect the presence of the Coronavirus in the lung based on a CT scan. The methods: Two global datasets are adopted. However, the weakness was that the datasets did not contain cases. Therefore, these two groups were merged. One Dataset was formed and used in the proposed model, which is based on deep learning and initial processing operations in a new way. Removing noise (buy Fast mean filter) and the second Filter redistributes colors (CLAHE Filter), as it opens dark areas and darkens light areas, then a filter is used to preserve the borders of shapes (Laplace). The penultimate stage is the training stage using (densenet-121) and reviewing the results through (confusion matrices). The proposed system obtained high-accuracy results of up to 97% in real terms and provided the system with interfaces that supported the user's work on it. Conclusions: The most critical step in establishing a medical system is choosing a global dataset to train on. In the proposed system, pruning and formatting the data was an essential step to conduct a robust training process. As the data was pruned and two files were created, including the injured and the other without the infected, the image processing operations carried out through three filters were very efficient due to the abundance of very high-resolution images free of noise. There is also a prediction through which a picture of the proposed system can be given, providing the user with the results of whether he is infected and the percentage of his injuries.
