Exploration Of Coronavirus (Covid-19) Outbreak, Empowered With Deep Learning Algorithms: Challenges And Research Opportunities

Globally, the coronavirus outbreak has had a significant impact on people and healthcare systems.The need for quick and efficient screening techniques to identify COVID-19 infections at an early stage is critical.Thirty (30) research articles that investigate the use of different deep learning models for COVID-19 infection identification and classification are carefully evaluated in this review.Using reliable scholarly databases, targeted keywords, relevance, publication dates ranging from 2019 to 2023, and language requirements, the review conducts a methodical literature search.The algorithms employed, their contributions, performance measures, and limits were taken into consideration when reviewing and analyzing each chosen study.A number of noteworthy restrictions were noted, such as the use of small COVID-19 sample datasets.As a result, augmentation and preprocessing techniques were used, including image sharing, zooming, rotating, resizing, and standardization, to reduce overfitting and increase dataset size.The results showed encouraging accuracy.Moreover, some studies solely focused on coronavirus binary classification that is, either positive or negative which limited the study of multi-class classification that might differentiate between other respiratory diseases, including bacterial pneumonia and other viral infections.Additionally, a thorough assessment was lacking in certain research.Because explainable AI (XAI) is still lacking, it is difficult to use these models in real-world healthcare systems.Robust, scalable, and flexible solutions are needed.This is because users must be able to comprehend and trust the outcomes.These discoveries are essential for subsequent investigations focused on creating more reliable and workable methods for the early identification and categorization of coronavirus diseases.