Table 3.

Performance of radiomics, clinicoradiological and combined radiomics–clinical models in prediction of residual uterine myoma regrowth after high‐intensity focused ultrasound ablation

Model	AUC	Sensitivity	Specificity	PPV	NPV	Accuracy
Training set (n = 243)
Radiomics	0.835 (0.783–0.886)	0.788 (0.709–0.852)	0.774 (0.680–0.847)	0.818 (0.740–0.878)	0.739 (0.645–0.815)	0.782 (0.726–0.829)
Clinicoradiological	0.911 (0.874–0.947)	0.829 (0.754–0.885)	0.845 (0.756–0.906)	0.879 (0.808–0.927)	0.784 (0.694–0.854)	0.835 (0.783–0.877)
Combination	0.933 (0.901–0.965)	0.894 (0.825–0.939)	0.874 (0.794–0.927)	0.894 (0.825–0.939)	0.874 (0.794–0.927)	0.885 (0.838–0.919)
Internal test set (n = 81)
Radiomics	0.834 (0.747–0.920)	0.740 (0.594–0.849)	0.742 (0.551–0.875)	0.822 (0.674–0.915)	0.639 (0.462–0.787)	0.741 (0.635–0.824)
Clinicoradiological	0.888 (0.816–0.960)	0.755 (0.608–0.862)	0.750 (0.563–0.879)	0.822 (0.674–0.915)	0.667 (0.489–0.809)	0.753 (0.649–0.835)
Combination	0.922 (0.857–0.987)	0.869 (0.731–0.946)	0.857 (0.689–0.946)	0.889 (0.752–0.958)	0.833 (0.665–0.930)	0.864 (0.771–0.924)
External test set (n = 104)
Radiomics	0.801 (0.712–0.889)	0.868 (0.711–0.951)	0.712 (0.586–0.814)	0.635 (0.489–0.760)	0.904 (0.782–0.964)	0.769 (0.679–0.840)
Clinicoradiological	0.912 (0.851–0.973)	0.767 (0.637–0.862)	0.864 (0.719–0.943)	0.885 (0.759–0.952)	0.731 (0.587–0.840)	0.808 (0.721–0.873)
Combination	0.930 (0.880–0.980)	0.891 (0.756–0.959)	0.810 (0.682–0.897)	0.789 (0.649–0.885)	0.904 (0.782–0.964)	0.846 (0.764–0.904)

Values in parentheses are 95% CI. AUC, area under the receiver‐operating‐characteristics curve; NPV, negative predictive value; PPV, positive predictive value.