Fig. 5: GDM discriminant model constructed using selected features from GDM biomarkers. | Nature Communications

Fig. 5: GDM discriminant model constructed using selected features from GDM biomarkers.

From: Metabolomics of saliva, serum, and urine for pathogenesis, diagnosis, and prognosis in gestational diabetes mellitus

Fig. 5

a Seven discriminant models were constructed using selected features from GDM biomarkers in saliva, serum, and urine. Their performances in the internal test set were shown. The optimal GDM discriminant model was constructed from multivariate samples with an AUC value of 0.868 (95%CI, 0.781–0.955), followed by the binary samples models (saliva+serum: 0.842 [0.744–0.940], saliva+urine: 0.836 [0.738–0.933], serum+urine: 0.861 [0.761–0.956]) and the single sample models (saliva: 0.773 [0.605–0.942], serum: 0.779 [0.670–0.888], urine: 0.747 [0.565–0.928]). b In the external test set, the AUC value of GDM discriminant model constructed with saliva, serum, and urine was 0.796 (95%CI, 0.695–0.897). Source data were provided as a Source Data file.

Back to article page