The Role of Artificial Neural Networks Techniques in the Fight Against COVID-19: A Review

This review presents an overview of the current public health emergency and the development of Artificial Neural Network techniques incorporation with medical imaging, such as chest computed tomography images and chest X-rays. Recent observations suggest that many medical researchers are increasingly employed modern computational methods to combat the spread of coronavirus disease 2019 (COVID-19) worldwide. These methods can be divided into six directions of focus: screening, detection, diagnosis, classification, drug repurposing, and disease prediction & forecasting. This research suggests critical reviews and summarizes key studies that analyze the techniques used in these six directions of combating COVID-19, which include hybrid Artificial Neural Network models, cross-dataset validation, and integrating these technologies into clinical practice. It provides a comprehensive survey of the effect of Artificial Neural Networks in healthcare during the COVID-19 pandemic. Various Artificial Neural Network techniques have been used to enhance and expedite COVID-19 diagnostics, ultimately aiming to reduce the severity of symptoms through early detection and assessment of infection rates. In the literature retrieval process for this review, 130 research papers were taken; from this collection of papers, 10 papers that did not contain useful information were excluded, and 120 papers very relevant to our review analytics were included. This research was divided according to its direction of work; the relevant studies were sourced from multiple citation databases, including Elsevier, ScienceDirect, Google Scholar, Springer, ResearchGate, and IEEE Preprints. The review concluded that among the 120 studies related to Covid-19, 62.5% utilized Convolutional Neural Network models, 9.2% employed Recurrent Neural Networks, 7.5% used Support Vector Machines, 3.3% applied Radial Basis Functions, and 17.5% implemented various other types of machine learning models.