Table 3 Performance of combined models in the validation group.

From: The value of intratumoral and peritumoral ultrasound radiomics model constructed using multiple machine learning algorithms for non-mass breast cancer

Combined modelsa

AUC (95% CI)

ACC (95% CI)

SEN (95% CI)

SPE (95% CI)

PPV (95% CI)

NPV (95% CI)

AUDC (95%CI)

All (peri_1mm_XGBoost)

0.799 (0.707–0.891)

0.740 (0.640–0.840)

0.889 (0.816–0.962)

0.610 (0.499–0.721)

0.667 (0.559–0.774)

0.862 (0.782–0.942)

0.104 (−0.274–0.251)

All (peri_2mm_RF)

0.850 (0.767–0.932)

0.805 (0.714–0.896)

0.882 (0.808–0.957)

0.744 (0.644–0.844)

0.732 (0.630–0.833)

0.889 (0.816–0.962)

0.294 (−0.530–0.549)

All (peri_3mm_XGBoost)

0.833 (0.748–0.919)

0.753 (0.655–0.852)

0.723 (0.621–0.826)

0.800 (0.708–0.892)

0.850 (0.768–0.932)

0.649 (0.540–0.757)

0.160 (−0.354–0.361)

All (peri_4mm_SVM)

0.782 (0.688–0.877)

0.740 (0.640–0.840)

0.771 (0.675–0.868)

0.714 (0.611–0.817)

0.692 (0.587–0.798)

0.789 (0.696–0.883)

0.160 (−0.292–0.262)

All (peri_5mm_LR)

0.830 (0.744–0.916)

0.805 (0.714–0.896)

0.769 (0.673–0.865)

0.842 (0.758–0.926)

0.833 (0.748–0.919)

0.780 (0.686–0.875)

0.205 (−0.323–0.294)

All (in_XGBoost)

0.858 (0.778–0.939)

0.792 (0.699–0.885)

0.756 (0.657–0.854)

0.844 (0.760–0.927)

0.872 (0.794–0.949)

0.711 (0.607–0.814)

0.282 (−0.306–0.297)

  1. aThe optimal radiomics models from each ROI group were combined with independent clinical and ultrasound predictive factors.
  2. AUC: area under curve, ACC: accuracy, SEN: sensitivity ,SPE: specificity ,PPV: positive predictive Value, NPV: negative predictive value, AUDC: area under the decision curve analysis, CI: confidence interval, RF: random forest, SVM: Support Vector Machine, LR: Logistic Regression.
Back to article page