Skip to main content

Table 2 Prediction performance and statistical comparisons of PFS among three models

From: Personalized prediction of immunotherapy response in lung cancer patients using advanced radiomics and deep learning

Estimate

Prediction time points

1 month

2 months

3 months

6 months

Model1: Clinical features (C-index = 0.66)

 AUC (95% CI a)

0.73 (0.73–0.93)

0.70 (0.64–0.82)

0.73 (0.65–0.86)

0.74 (0.66–0.89)

 Sensitivity

68%

67%

69%

69%

 Specificity

68%

69%

67%

69%

Model2: Clinical + intratumoral (C-index = 0.68)

 AUC (95% CI)

0.71 (0.60–0.75)

0.74 (0.64–0.79)

0.77 (0.68–0.82)

0.68 (0.67–0.70)

 Sensitivity

73%

72%

71%

71%

 Specificity

67%

67%

74%

66%

Model3: Clinical + intratumoral + peritumoral-vasculature (C-index = 0.83)

 AUC (95% CI)

0.80 (0.71–0.93)

0.76 (0.72–0.84)

0.80 (0.71–0.88)

0.77 (0.70–0.95)

 Sensitivity

70%

75%

76%

82%

 Specificity

74%

70%

76%

70%

Statistical comparison in AUC (p-value)

 Model 1 vs. 3

< 0.001

< 0.001

< 0.001

< 0.001

 Model 2 vs. 3

< 0.001

0.004

< 0.001

< 0.001

  1. a CI: Confidence interval in 100 times bootstrap sampling