Skip to main content
Fig. 4 | BMC Infectious Diseases

Fig. 4

From: Projecting COVID-19 disease severity in cancer patients using purposefully-designed machine learning

Fig. 4

Important clinical variables identified by the model. a The top 30 variables (y-axis) and their importance (x-axis), defined using permutation testing. The category of each variable is listed next to its name: B = Basic patient information, C = Cancer-related, I = ICD codes, R = Radiology, L = Laboratory. b The performance of the classifier (y-axis) when trained using variables from each category separately. For example, using only radiology variables, the random forest classifier achieved an AUROC, averaged over all three classes, of 60.6%. “All” shows the combination of all variables, achieving an average AUROC of 74.7%

Back to article page