An effective hybrid attention capsule autoencoder model for diagnosing COVID-19 disease using chest CT scan images in an edge computing environment

Due to the severe COVID-19 pandemic, securing and processing patient's health data for decision-making is challenging. The proposed study introduces an edge computing-assisted hybrid deep learning mechanism to mitigate this issue. Initially, the first layer in a proposed framework collects the input images and forwards them to the edge layer. Here, several edge devices are deployed, and COVID-19 disease detection is performed through a new hybrid attention capsule autoencoder (HA_CAE). Finally, the samples are securely stored in the cloud layer, in which the reliability of patient information is increased. The simulation analysis shows that the proposed model gains higher detection performance, like accuracy (99.1%), precision (97.6%), sensitivity (98.9%), specificity (96.8%), F1-score (98.2%), and MCC (97.6%).