Skip to main content

Table 4 Model utility based on cutoff values in the complete case analysis of the training set

From: Atypical ductal hyperplasia diagnosed by US-guided core needle biopsy: clinical, pathological and US features associated with upgrading to malignancy

Groups

Sensitivity

Specificity

Upgrade rates

Low-risk group

High-risk group

Training set

    

 0.63a

81.4(70/86)

86.7(52/60)

23.5(16/68)

89.7(70/78)

 0.18a

95.3(82/86)

56.7(34/60)

10.5(4/38)

75.9(82/108)

 0.11a

98.8(85/86)

38.3(23/60)

4.1(1/24)

69.7(85/122)

Internal validation set

    

 0.63a

77.8(21/27)

90.0(18/20)

25.0(6/24)

91.3(21/23)

 0.18a

100(27/27)

50.0(10/20)

0

73.0(27/37)

 0.11a

100(27/27)

35.0(7/20)

0

67.5(27/40)

External validation set

    

 0.63a

73.1(19/26)

90.3(28/31)

20.0(7/35)

86.4(19/22)

 0.18a

100(26/26)

67.7(21/31)

0

72.2(26/36)

 0.11a

100(26/26)

41.9(13/31)

0

59.1(26/44)

  1. Note: variables are expressed as numbers of lesions with percentages in parentheses
  2. a Cutoff values obtained from the training set