Research Paper Volume 13, Issue 6 pp 8929—8943

Blood metabolomic profiling predicts postoperative gastrointestinal function of colorectal surgical patients under the guidance of goal-directed fluid therapy

Support vector machine (SVM) analysis of metabolites in blood samples in the PACU and at 24 h after surgery in patients with delayed and enhanced postoperative recovery. (A) SVM classification of samples in groups (A, B). (B) Specificity, sensitivity, and predictive accuracies of the (A, B) SVM model. (C) Features’ importance ranking of the (A, B) SVM model. (D) SVM classification of samples belonging to groups (C, D). (E) Specificity, sensitivity, and predictive accuracies of the (C, D) SVM model. (F) Features’ importance ranking of the (C, D) SVM model. The red cubes represent higher importance, and the green cubes represent lower importance in the SVM model.

Figure 4. Support vector machine (SVM) analysis of metabolites in blood samples in the PACU and at 24 h after surgery in patients with delayed and enhanced postoperative recovery. (A) SVM classification of samples in groups (A, B). (B) Specificity, sensitivity, and predictive accuracies of the (A, B) SVM model. (C) Features’ importance ranking of the (A, B) SVM model. (D) SVM classification of samples belonging to groups (C, D). (E) Specificity, sensitivity, and predictive accuracies of the (C, D) SVM model. (F) Features’ importance ranking of the (C, D) SVM model. The red cubes represent higher importance, and the green cubes represent lower importance in the SVM model.