Skip to main content

Table 3 Performance evaluation of the radiomic models using SVM algorithm in the training and validation cohort

From: Predicting response to immunotherapy plus chemotherapy in patients with esophageal squamous cell carcinoma using non-invasive Radiomic biomarkers

  Models Accuracy Sensitivity Specificity NPV PPV AUC
Training cohort 3D uncorrected 0.701 0.590 0.814 0.720 0.734 0.626
(0.690–0.718) (0.570–0.622) (0.796–0.831) (0.700–0.735) (0.702–0.754) (0.602–0.637)
3D corrected 0.690 0.581 0.814 0.705 0.752 0.628
(0.680–0.702) (0.556–0.607) (0.792–0.834) (0.694–0.721) (0.720–0.776) (0.583–0.611)
2D uncorrected 0.801 0.693 0.900 0.779 0.915 0.776
(0.800–0.821) (0.681–0.715) (0.886–0.932) (0.771–0.799) (0.910–0.932) (0.772–0.791)
2D corrected 0.804 0.727 0.886 0.795 0.917 0.818
(0.793–0.815) (0.706–0.742) (0.855–0.900) (0.784–0.803) (0.896–0.925) (0.797–0.829)
Validation cohort 3D uncorrected 0.640 0.431 0.864 0.602 0.750 0.531
(0.632–0.666) (0.36–0.49) (0.813–0.900) (0.575–0.631) (0.694–0.811) (0.502–0.560)
3D corrected 0.640 0.432 0.861 0.601 0.750 0.514
(0.631–0.660) (0.363–0.491) (0.800–0.911) (0.570–0.632) (0.691–0.811) (0.480–0.544)
2D uncorrected 0.790 0.709 0.860 0.710 0.852 0.729
(0.770–0.801) (0.681–0.756) (0.830–0.891) (0.564–1.000) (0.830–0.881) (0.711–0.760)
2D corrected 0.796 0.714 0.872 0.753 0.848 0.787
(0.770–0.806) (0.673–0.767) (0.841–0.901) (0.721–0.786) (0.813–0.875) (0.752–0.806)
  1. Abbreviations: SVM-Support Vector Machine, AUC-Area under the Receiver-Operating Characteristic Curve, NPV-Negative Predictive Value, PPV-Positive Predictive Value