Consume Machine Learning Integration Algorithms to Predict COVID-19 Infection

ABSTRACT

In the context of the emerging coronavirus pneumonia epidemic becoming a global epidemic, nucleic acid testing as a as a precise prevention and control method has been universally recognized, but because the scope of the test is too big and the production process is complicated, the kits produced by biological companies are difficult to use widely, for this reason I develop some machine learning integrated algorithms which can forecast whether a man is infected with COVID-19 based on three highly accessible features. This method can predict whether a person has been infected with COVID-19 based only on three indicators heart rate, blood oxygen level, and body surface temperature, and we use several tree integration. We used several tree integration algorithms such as Random Forest, XGBoost, and GBM, and its accuracy, recall, and F1 score obtained 100% accuracy on the test set, which has been better than the current nucleic acid detection methods, proving that this method can be theoretically used as an accurate, convenient, and efficient self-detection method. © 2022 IEEE.