Fig. 5.

ROC-analysis of serum trace elements to predict survival of patients with COVID-19. (A) Serum concentrations of Se (pale grey), Zn (grey) and SELENOP (black) predict survival of patients with COVID-19 with similar and moderately good precision. (B) The sums of serum Zn with serum Se or serum Zn with serum SELENOP concentrations both yield a higher area under the curve (AUC). (C) The multiple regression model based on Zn, SELENOP and the patient's age outperformed any other combination of variables via stepwise AIC selection. The final model, based on these three parameters, yielded the highest AUC of 94.42%. The corresponding cutpoint, according to the Youden Index [48] is indicated.