Figure 6. Random forest to detect signatures of severity in the SARS-CoV-2-infected subjects.

A. Receiver operating characteristic (ROC) curve of random forest classification on patient intubation status using OP and NP samples. Top 15 most important predictors in classifying patient’s intubation status using OP (B) and NP (C) samples are shown. D. ROC curve of random forest classification on disease severity using both OP and NP samples. Top 15 most important predictors in classifying disease severity using OP (E) and NP (F) samples are shown.