Machine Learning Algorithms in Analysis, Diagnosing and Predicting COVID-19: A Systematic Literature Review

Since the COVID-19 corona virus first appeared at the end of 2019, in Wuhan province, China, the analysis, diagnosis, and prognosis of COVID-19 (SARS-CoV-2) has attracted the greatest attention. Since then, every part of the world needs some sort of system or instrument to assist judgments for prompt quarantine and medical treatment. For a variety of uses, including prediction, classification, and analysis, machine learning (MLR) have demonstrated their accuracy and efficiency in the fields of education, health, and security. In this paper, three main questions will be answered related to COVID-19 analysis, predicting, and diagnosing. The performance evaluation, fast process and identification, quick learning, and accurate results of MLR algorithms make them as a base for all models in analyzing, diagnosing, and predicting COVID-19 infection. The impact of using supervised and unsupervised MLR can be used for estimating the spread level of COVID-19 to make the proper strategic decisions. The researchers next compared the effects of various datatypes on diagnosing, forecasting, and assessing the severity of COVID-19 infection in order to examine the effects of MLRs. Three fields are associated with COVID-19, according to the analysis of the chosen study (analysis, diagnosing, and predicting). The majority of researches focus on the subject of COVID-19 diagnosis, where they use their models to identify the infection. In the selected studies, several algorithms are employed, however, a study revealed that the neural network is the most used method when compared to other algorithms. The most used method for identifying, forecasting, and evaluating COVID-19 infection is supervised MLR