Severity Detection For the Coronavirus Disease 2019 (COVID-19) Patients Using a Machine Learning Model Based on the Blood and Urine Tests

Authors:

Abstract
Background: The recent outbreak of the coronavirus disease-2019 (COVID-19) caused serious challenges to the human society in China and across the world. COVID-19 induced pneumonia in human hosts and carried a highly inter-person contagiousness. The COVID-19 patients may carry severe symptoms, and some of them may even die of major organ failures. We aim to utilize the machine learning algorithms to build the COVID-19 severeness detection model. Methods: This study recruited the binary classification problem between 75 severely illed COVID-19 infected patients and the other 62 patients with mild symptoms. Support vector machine (SVM) demonstrated a promising detection accuracy after 32 features were detected to be significantly associated with the COVID-19 severeness. These 32 features were further screened for inter-feature redundancies. Findings: The severely illed patients had a higher serum level of neutrophil percentage and lower serum levels of monocyte percentage and calcium compared with those mild ones. The blood test features demonstrated much more significant inter-group differences than the urine test features. These three blood test features as candidate severeness biomarkers, i.e., serum ferritin, hs-CRP, interleukin-2R, and tumor necrosis factor-α. The final SVM model achieved the overall accuracy 0.8148 using 28 features. Interpretation: This study utilized the machine learning algorithms to detect the COVID-19 severely ill patients from those with only mild symptoms. Our experimental data demonstrated strong correlations with the COVID-19 severeness. The 28 biomarkers may also be investigated for their underlining mechanisms of their roles in the COVID-19 severely ill patients. Funding Statement: This work was supported by grants from The epidemiology, early warning and response techniques of major infectious diseases in the Belt and Road Initiative (#2018ZX10101002), National Natural Science Foundation of China (#81871699), Jilin Provincial Key Laboratory of Big Data Intelligent Computing (20180622002JC), the Education Department of Jilin Province (JJKH20180145KJ), Foundation of Jilin Province Science and Technology Department (#172408GH010234983), and the startup grant of the Jilin University. This work was also partially supported by the Bioknow MedAI Institute (BMCPP-2018-001), the High Performance Computing Center of Jilin University, and the Fundamental Research Funds for the Central Universities, JLU. Declaration of Interests: The authors declare no competing interests. Ethics Approval Statement: This study was approved by the Ethics Commission of the First Hospital of Jilin University. Informed consent was waived for this emerging infectious disease.