Fig. 4: The performance of the AI system in identifying COVID-19 pneumonia from CXR images.

a,b,d,e, ROC curves (a,d) and normalized confusion matrices (b,e) for binary classification. a,b, The performance of the AI system in differentiating between COVID-19 pneumonia and other pneumonia (‘Others’, for example, bacterial pneumonia) on the test dataset: AUC = 0.966 (95% CI = 0.955–0.975), sensitivity = 92.07%, specificity = 90.12%. d,e, The performance of the AI system in differentiating between COVID-19 pneumonia and other viral pneumonia (OVP) on the test dataset: AUC = 0.867 (95% CI = 0.828–0.902), sensitivity = 82.32%, specificity = 72.63%. c,f, ROC curves showing the performance of the AI system in identifying severe or non-severe COVID-19 from other pneumonia (c) and other types of viral pneumonia (f).