A Systematic Review and Analysis on Deep Learning Techniques Used in Diagnosis of Covid-19

Abstract

Contagious diseases are very common and dangerous. To find out if they are infected, people must go to doctors directly or indirectly. Coronavirus disease and its different forms have had a big effect on the world in the last few years. In this work, a deep learning model is put into place that can use cough sounds to find COVID-19. You can find out if someone has a disease without having to touch them. For this study, more than 60 research articles are considered from various indexing sources. After several selection steps, finally a 31 quality research articles are included for detailed analysis. From this study, it is observed that majority of the research articles focused on DL techniques with cough audio signals. The classification of cough sounds helps doctors figure out what's way off the mark with the respiratory system. The COUGHVID dataset, which has cough sound files, is used. The sound is turned into spectrogram images, and then an auto encoder with an attention mechanism is used to pull out the features. The features are then fed into an algorithm that uses deep learning to figure out if a person is COVID positive or not. This model is utilized by either the medical professionals or the patients in order to do the initial disease diagnosis.