Published Paper Details:

An unexpected pandemic known as COVID-19 struck the entire world in the year 2020. The lack of a cure has spurred research into all fields to address it. The discovery of techniques for the diagnosis, detection, and prediction of COVID-19 instances is a major contribution to computer science. The most popular methods in this field are data science and machine learning (ML). This document provides a summary of over 160 ML-based strategies created to tackle COVID-19. These sources include Elsevier, Springer, ArXiv, MedRxiv, and IEEE Xplore, among others. They are examined and divided into two groups: techniques based on supervised learning and those based on deep learning. The ML algorithm used in each category is described, along with a list of the parameters that were applied. The settings for each algorithm's parameters are compiled in several tables. They comprise the nature of the examined data (text data, X-ray pictures, CT images, time series, clinical data, etc.), the type of the addressed problem (detection, diagnosis, or detection), and the evaluated metrics (accuracy, precision, sensitivity, specificity, F1-Score, and AUC). The study explores the data gathered and offers some statistics that paint a picture of the current state of the art. According to the results, 79% of cases include deep learning, 65% of which are based on convolutional neural networks (CNN), and 17% employ specialised CNN. On the other hand, only Random Forest, (SVM), and Regression algorithms are used, and supervised learning is only present in 16% of the approaches that were assessed.