This paper aims to explore the role of Machine Learning (ML) techniques in combating COVID-19. In this study, different ML techniques and data portioning methods will be used to predict RT-PCR results from ER-admitted patients. The data set contains 199 instances with 81 attributes. 5-Fold Cross Validation, 10-Fold Cross Validation, and 80% Training are the different data portioning methods utilized for this research. Decision tree (J48), Random Forest (RaF), and Rotation Forest (RoF), Multi-Layer Perceptron (MLP), Naïve Bayes (NB), K-Nearest Neighbors (kNN), Logistic Regression (LR), LogitBoost (LB), and Sequential Minimal Optimization (SMO) are the main classifiers we explore in this study. The results of our experiments indicate that Rotation Forest gives a highest accuracy of 90% on the data set.

MOTIVATION: The development of new high-throughput genotyping products requires a significant investment in testing and training samples to evaluate and optimize the product before it can be used reliably on new samples. One reason for this is current methods for automated calling of genotypes are based on clustering approaches which require a large number of samples to be analyzed simultaneously, or an extensive training dataset to seed clusters. In systems where inbred samples are of primary interest, current clustering approaches perform poorly due to the inability to clearly identify a heterozygote cluster. RESULTS: As part of the development of two custom single nucleotide polymorphism genotyping products for Oryza sativa (domestic rice), we have developed a new genotype calling algorithm called 'ALCHEMY' based on statistical modeling of the raw intensity data rather than modelless clustering. A novel feature of the model is the ability to estimate and incorporate inbreeding information on a per sample basis allowing accurate genotyping of both inbred and heterozygous samples even when analyzed simultaneously. Since clustering is not used explicitly, ALCHEMY performs well on small sample sizes with accuracy exceeding 99% with as few as 18 samples. AVAILABILITY: ALCHEMY is available for both commercial and academic use free of charge and distributed under the GNU General Public License at http://alchemy.sourceforge.net/ CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.