Multicenter analysis and a rapid screening model to predict early novel coronavirus pneumonia using a random forest algorithm

Abstract

Early determination of coronavirus disease 2019 (COVID-19) pneumonia from numerous suspected cases is critical for the early isolation and treatment of patients. The purpose of the study was to develop and validate a rapid screening model to predict early COVID-19 pneumonia from suspected cases using a random forest algorithm in China. A total of 914 initially suspected COVID-19 pneumonia in multiple centers were prospectively included. The computer-assisted embedding method was used to screen the variables. The random forest algorithm was adopted to build a rapid screening model based on the training set. The screening model was evaluated by the confusion matrix and receiver operating characteristic (ROC) analysis in the validation. The rapid screening model was set up based on 4 epidemiological features, 3 clinical manifestations, decreased white blood cell count and lymphocytes, and imaging changes on chest X-ray or computed tomography. The area under the ROC curve was 0.956, and the model had a sensitivity of 83.82% and a specificity of 89.57%. The confusion matrix revealed that the prospective screening model had an accuracy of 87.0% for predicting early COVID-19 pneumonia. Here, we developed and validated a rapid screening model that could predict early COVID-19 pneumonia with high sensitivity and specificity. The use of this model to screen for COVID-19 pneumonia have epidemiological and clinical significance.