RBTM: Prediction of COVID-19 from Random Blood Test with Machine Learning Algorithms

ABSTRACT

Recently, the ongoing global pandemic of novel coronavirus infection had a devastating impact worldwide. We develop an efficient classification model that effectively produces the predictive values of infected patients with suspicious symptoms and epidemiological history to defeat this. The research aims to use the Traditional technique to compare clinical blood tests of positive and negative cases. The diagnostic Machine Learning model incorporates 551random blood samples with the following parameters of the patient's demographic features, Platelet, Hemoglobin, Lymphocyte, Neutrophil, Leukocyte (WBC), Turbidimetric, Troponin-I of COVID positive and negative cases. The prediction model can achieve the classification report of Accuracy, Precision, Recall, and F1 score values. In this analysis, considering seven different algorithms for the prediction and the observation's estimation, the data is 5-fold cross-validated. Finally, investigational outcomes attain accurate predictions. Logistic Regression predicted 0.83% of accuracy. The Receiver Operator Characteristic (ROC) metrics for Logistic Regression, the Precision was 0.78%, Recall was 0.85%, and F1-score was 0.82%, Specificity was 0.58%, and Sensitivity was 0.41%. © 2022 IEEE.