A novel machine learning–based detection and diagnosis model for coronavirus disease (COVID-19) using discrete wavelet transform with rough neural network

Abstract Coronavirus disease-2019 (COVID-19) screening testing of a massive number of suspected cases for proper quarantine and treatment is a priority. Pathogenic laboratory testing is used as a diagnosis process, but it consumes much time with a high false detection rate. The development of artificial intelligence techniques has a crucial part in streamlining and accelerating the diagnosis of COVID-19 patients. To meet current requirements for COVID-19 diagnosis, it is highly important to develop an automated diagnosis model to identify the disease accurately and at a faster rate. According to COVID-19 radiographical changes in computed tomography images, this research designed a machine learning–based diagnosis model using discrete wavelet transform (DWT) with a rough neural network (RNN), called a DWT-RNN model. Moreover, principal component analysis was conducted to reduce the subset of features before classification. The DWT-RNN model was validated using a COVID-19 chest X-ray dataset. The experimental outcome was validated for several aspects and the obtained results show that the DWT-RNN model offers maximum classification performance and saves time in controlling diseases.