Rapid and accurate identification of COVID-19 infection through machine learning based on clinical available blood test results

Jiangpeng Wu; Pengyi Zhang; Liting Zhang; Wenbo Meng; Junfeng Li; Chongxiang Tong; Yonghong Li; Jing Cai; Zengwei Yang; Jinhong Zhu; Meie Zhao; Huirong Huang; Xiaodong Xie; Shuyan Li

Coronavirus disease 2019 (COVID-19) 1 can induce emerging acute respiratory illness by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) which is classified by International Committee on Taxonomy of Viruses (ICTV). 2 This enveloped RNA virus is the seventh member of the coronavirus family to infect humans belonging to lineage B of the genus betacoronavirus through deep sequencing and phylogenetic analysis. 3 The epidemic pneumonia has spread quickly and become a global public health concern. 4 After making careful assessment for COVID-19, the epidemic is classified as a pandemic due to the alarming spread of the infection and the severity of the disease in March 11, 2020 . 5 A large amount of practice and research show that the common clinical symptoms of the pneumonia are fever, cough, myalgia or fatigue. 6 And the characteristics of chest radiography or computed tomography (CT) are multiple mottling or ground-glass opacity. 7 According to these valuable experience and achievements, the seventh edition of a new coronavirus pneumonia diagnosis and treatment plan has jointly updated by two Chinese departments. The official guidelines emphasize that the use of quantitative real-time PCR (qRT-PCR) assays for the detection of viral nucleic acid to diagnose the infected patients is still one of the main reference standards. 8 Although RNA of this coronavirus is the prime marker that can be detected in the early stage of infection, the high requirements of its testing technology and condition limit its positive rate to as about 30-50%. 9 This situation may imply that many patients with COVID-19 cannot be identified in time. 10 The situation may delay the isolation and treatment of infected patients and accelerate the spread of the epidemic due to the characteristic of human-to-human transmission. 11 Apart from this, some objective factors, such as complicated operation, time-consuming, easy to pollute, high cost and so on, further hinder the technology from exerting its due advantages on emergency prevention and control conditions. 12 Many clinicians strongly recommended the use of CT imaging in severe epidemic areas, however, the tardy method would present obvious abnormalities when the lung produced inflammatory or the tissue generated lesions. 13 At present, the combined detection of immunoglobulin M (IgM) and immunoglobulin G (IgG) antibody provides a fast and simple screening method, but the two substances are impossible to be detected in the window period of about two weeks. 14, 15 So this method can only be used as an effective supplement and auxiliary diagnosis of nucleic acid detection. 16 Until now, the quick diagnosis of COVID-19 patients in a large scale is still a major problem for global health concern. 17 Except for the laboratory-confirmed COVID-19 patients, there are a large number of suspected patients and close contacts who need to be confirmed whether they are infected or not. 18 In this case, it is difficult to screen and identify the cases during the incubation period or subclinical infections, they have great potential to become super spreaders. 19 Although timely isolation measures can effectively cut down the spread of the virus, early diagnosis can enable patients to get more active treatment at the initial stage of presentation. 20 Therefore, the early identification and timely management of COVID-19 patients is of great importance due to no specific treatment. 21 . CC-BY-NC-ND 4.0 International license I