Influenza and Anosmia: Important Prediction Factors for Severity and Death of COVID-19

Abstract

Abstract Many people are being hospitalized and are dying from COVID-19 as the right treatment has not yet been identified. We investigate the factors related to the severity and mortality of COVID-19 using big data-machine learning techniques. This retrospective study included 8070 SARS-CoV-2 confirmed patients of the 129,120 SARS-CoV-2 RNA tested patients in South Korea between January and July 2020, and whose data were available from the National Health Insurance Service. Primary endpoint was comorbidity, severity and mortality rate in COVID-19. Machine learning algorithms were performed to evaluate the effects of comorbidities on severity and mortality rate of COVID-19. The most common comorbidities of COVID-19 were pulmonary inflammation followed by anosmia. The model that best predicted severity was a neural network (AUC: 85.06%). The most important variable for predicting severity in the neural network model was a history of influenza (relative importance: 0.129). The model that best predicted mortality was the logistic regression elastic net (EN) model (AUC: 93.86%). The most important variables for mortality in the EN model were age (coefficient: 2.136) and anosmia (coefficient: –1.438). Through the state-of-the-art machine learning algorithm and 8070 patients of COVID-19 patients in South Korea, influenza was found to be a major adverse factor in addition to old age and male sex. In addition, anosmia was found to be a major factor associated with lower severity and mortality rates. The patient’s history of influenza and anosmia will be an important indicator for predicting the severity and mortality of COVID-19 patients.