COVID-19 diagnosis by routine blood tests using machine learning

Abstract

Physicians taking care of patients with COVID-19 have described different changes in routine blood parameters. However, these changes hinder them from performing COVID-19 diagnoses. We constructed a machine learning model for COVID-19 diagnosis that was based and cross-validated on the routine blood tests of 5333 patients with various bacterial and viral infections, and 160 COVID-19-positive patients. We selected the operational ROC point at a sensitivity of 81.9% and a specificity of 97.9%. The cross-validated AUC was 0.97. The five most useful routine blood parameters for COVID-19 diagnosis according to the feature importance scoring of the XGBoost algorithm were: MCHC, eosinophil count, albumin, INR, and prothrombin activity percentage. t-SNE visualization showed that the blood parameters of the patients with a severe COVID-19 course are more like the parameters of a bacterial than a viral infection. The reported diagnostic accuracy is at least comparable and probably complementary to RT-PCR and chest CT studies. Patients with fever, cough, myalgia, and other symptoms can now have initial routine blood tests assessed by our diagnostic tool. All patients with a positive COVID-19 prediction would then undergo standard RT-PCR studies to confirm the diagnosis. We believe that our results represent a significant contribution to improvements in COVID-19 diagnosis.

COVID-19 diagnosis from routine blood tests using artificial intelligence techniques

“…Following the December of 2019, the SARS-CoV-2 first reported in Wuhan, China [1] , [4] , [5] , [6] , [7] , [9] , [10] caused a pandemic (declared by World Health Organization (WHO) on March 11th, 2020) [1] , [5] , [9] by inducing respiratory infection (called COVID-19) with symptoms typical of fever, tiredness, and coughs [5] , [7] , [9] . While being highly contagious [1] , in some cases, COVID-19 infection could be asymptomatic making it capable of spreading at an increasing pace [2] , [6] , [7] . This poses a challenge to most of the countries worldwide with a much more burden on developing and less-developed countries.…”

“…Naturally, developed countries have faster and more comprehensive access to vaccines, while other countries are facing multiple hindrances progressing in vaccination course of action like shortage of sufficient vaccine doses for the vaccination of vulnerable groups. Moreover, there are still no confirmed medications to cure patients infected with COVID-19 [1] . Thus, the importance of screening patients suspected to be infected with the SARS-CoV-2 has not declined [1] .…”

“…Moreover, there are still no confirmed medications to cure patients infected with COVID-19 [1] . Thus, the importance of screening patients suspected to be infected with the SARS-CoV-2 has not declined [1] .…”

Time series prediction of COVID-19 transmission in America using LSTM and XGBoost algorithms

“…As the number of cases increases and more data becomes available, various researches [1] , [2] , [3] , [4] , [5] , [6] , [7] develop a range of mathematical models or employ machine learning algorithms to forecast the transmission of SARS-CoV-2. Previous studies have also employed LSTM [8] , [9] , [10] , [11] , [12] or XGBoost [13] , [14] , [15] , [16] , [17] , [18] , [19] models to forecast the spread of COVID-19 and identify the most influential COVID-19 indicators. Chimmua et al [8] adopted LSTM algorithm to forecast confirmed cases in Canada within next two weeks and emphasized the significant role of social distance regular.…”

“…To quantify the impact of the COVID-19 pandemic on driving behavior, the authors [18] utilized explanatory XGBoost feature importance to evaluate the influence of COVID-19 and used seasonal ARIMA models to model. Kukar et al [19] used random forest, deep neural networks, and XGBoost algorithms to build models that predicted COVID-19 diagnosis based on regular blood test results, age, and gender. To the best of our knowledge, few studies have used the combined LSTM networks and XGBoost models to predict infectious diseases in time series analysis.…”