Figure 4. ROC curves based on the binary logistic regression model by the combination of three plasma metabolites (C8-ceramide, sphingosine and glutamine), and their prediction plots based on the optimal cutoff value from ROC curves.

(A) The HAPE samples from the discovery set were applied to construct a binary logistic regression model based on the combination of plasma C8-ceramide, sphingosine and glutamine, and the ROC curves of the discovery set ((A), left) and validation set ((A), right) were obtained from the above established prediction model. (B) The optimal cutoff value with the highest sensitivity and specificity in the ROC curves of the training set was obtained (0.4988) and applied to evaluate the prediction capacity (92.8% for discovery set ((B), left) and 84.1% for validation set ((B), right)) of the current model, where 0 and 1 on the x axis represent healthy controls and HAPE patients, respectively, and blue circle represent samples.