Artificial Intelligence Application in COVID-19 Diagnosis and Prediction

Background: On December 8, 2019, the first new coronavirus case was discovered in Wuhan, China, and an intensive outbreak incepted in the next month (about January 20). Virologicalists and epidemiologists predict that it will reach a peak in about 90 days and fade away till the end in about 4 months (Early April), the entire epidemic will terminate in early May. The daily rise in confirmed cases and the increase in the number of outbreak communities on the epidemic map always hit the nerves of panic. On January 31, 2020, the World Health Organization (WHO) declared China ’s new coronavirus epidemic an “public health emergency of international concern” (PHEIC). At this time, citizens from multiple nations including China express their grave concerns on the diagnosis, prediction and heal of the virus contagion. Objective: This research will analyze the COVID-19 general diagnosis index with artifical intelligence (AI) to improve the diagnosis accuracy for clinical purpose. Methods: We included 32 cases of COVID-19 diagnosed, and 85 undiagnosis patients in Taizhou Public Hygiene in Taizhou Hospital, of Zhejiang province, between January 17, 2020 and February 1, 2020, and the positive confirmed patients were diagnosized by real-time RT-PCR. At the same time, 85 patients with negative nucleic acid of COVID-19 were collected during the same period, and the results were followed up to February 2, 2020. Among radiological characteristics and laboratory data (the first result of the Group) for analysis. We used four types of AI techonlogy to screen important index regarding COVID-19 diagnosis. several classical and state-of-the-art attribute reduction and feature selection methods, sparse rescaled linear square regression (SRLSR), evolutionary non-dominated radial slots based algorithm, attribute reduction with multi-objective decomposition-ensemble optimizer (ARMED), Gradient boosted feature selection (GFS), recursive feature elimination (RFE), were applied to deal with this problem. Results: We employed four types of AI technology to screen all patients, achieve 18 indexes associated with significant COVID-19 diagnosis and the most important attribute is WBC, Eosinophil count, Eosinophil rate, 2019 novel coronavirus RNA (2019n-CoV) and Amyloid-A in laboratory, also matched with 2019 China virus diagnosis clinical guide. Conclusion: We developed a novel and accurate method to quickly achieve COVID-19 diagnosis association indexes to improve confirmed diagnosis rate for clinical use. Funding Statement: This work is funded by the Science and Technology Development Fund of Macau (FDCT/131/2016/A3, FDCT/0015/2018/A1, FDCT/126/2014/A3) and Start-up Research Grant (SRG2016-00082-FHS), the Multi-Year Research Grant (MYRG2019-00069-FHS, MYRG2016-00069-FST), the intramural research program of Faculty of Health Sciences, University of Macau, Guangzhou Science and Technology Innovation and Development of Special Funds, Grant no. EF003/FST-FSJ/2019/GSTIC, and Grant no. EF004/FST-FSJ/2019/GSTI, project of "New coronavirus infection and prevention" Emergency Scientific Research of Chongqing Education Commission of China, and the project of science and technology research program of Chongqing Education Commission of China(KJQN201903601). Declaration of Interests: None. Ethics Approval Statement: This study was approved by the Institutional Medical Ethics Review Board of Taizhou Hospital of Zhejiang Province.