Skip to main content

Table 2 Performance of different methods in original cohort

From: Developing a practical neurodevelopmental prediction model for targeting high-risk very preterm infants during visit after NICU: a retrospective national longitudinal cohort study

Balanced dataset (10-CV)

Independent test

Method

Variables numbers

ACC (%)

AUC (95% CIs)

ACC (%)

AUC (95% CIs)

CDelay: BSIDIII cognitive score < 85 at 24 months CA

 All-feature-in

  RF

83

66.0

0.68 (0.62–0.74)

76.7

0.71 (0.65–0.76)

  Lasso

72

61.1

0.64 (0.59–0.69)

65.1

0.69 (0.63–0.74)

 Coarse-to-fine features selection

  LR

29

65.6

0.70 (0.66–0.74)

69.5

0.72 (0.68–0.76)

  RF

29

61.1

0.66 (0.62–0.70)

73.0

0.67 (0.63–0.71)

  SVM

29

66.0

0.70 (0.64–0.74)

66.9

0.72 (0.67–0.77)

  EL-NDI

10

71.9

0.74 (0.71–0.77)

77.6

0.75 (0.72–0.78)

  EL-LR

10

67.1

0.72 (0.66–0.78)

70.7

0.75 (0.70–0.80)

MDelay: BSIDIII motor score < 85 at 24 months CA

 All-feature-in

  RF

83

72.7

0.76 (0.70–0.82)

73.0

0.71 (0.66–0.77)

  Lasso

77

66.7

0.73 (0.65–0.81)

64.6

0.69 (0.64–0.74)

 Coarse-to-fine features selection

  LR

29

70.3

0.77 (0.71–0.83)

69.5

0.71 (0.67–0.75)

  RF

29

70.5

0.74 (0.69–0.79)

71.6

0.69 (0.66–0.72)

  SVM

29

70.9

0.76 (0.71–0.81)

75.2

0.70 (0.66–0.74)

  EL-NDI

4

72.4

0.78 (0.74–0.82)

75.4

0.73 (0.70–0.76)

  EL-LR

4

71.7

0.77 (0.72–0.83)

72.2

0.73 (0.68–0.78)

CRegres: BSIDIII cognitive score declines≧ 15 between 6 and 24 months CA

 All-feature-in

  RF

75

71.8

0.78 (0.73–0.82)

71.7

0.78 (0.74–0.82)

  Lasso

67

71.2

0.79 (0.75–0.83)

70.1

0.79 (0.76–0.83)

 Coarse-to-fine features selection

  LR

29

72.9

0.78 (0.75–0.81)

71.6

0.80 (0.77–0.83)

  RF

29

70.1

0.75 (0.72–0.78)

70.6

0.74 (0.71–0.77)

  SVM

29

70.6

0.78 (0.74–0.82)

69.1

0.77 (0.73–0.80)

  EL-NDI

4

72.0

0.79 (0.76–0.82)

73.0

0.81 (0.77–0.85)

  EL-LR

4

71.0

0.78 (0.75–0.81)

70.8

0.80 (0.77–0.83)

MRegres: BSIDIII motor score declines ≧ 15 between 6 and 24 months CA

 All-feature-in

  RF

75

75.4

0.79 (0.76–0.82)

81.8

0.86 (0.83–0.90)

  Lasso

64

71.0

0.79 (0.76–0.82)

73.3

0.82 (0.78–0.86)

 Coarse-to-fine features selection

  LR

29

70.1

0.78 (0.74–0.82)

73.0

0.81 (0.77–0.85)

  RF

29

69.7

0.76 (0.73–0.79)

69.9

0.76 (0.73–0.79)

  SVM

29

71.6

0.79 (0.76–0.82)

74.1

0.81 (0.78–0.84)

  EL-NDI

4

74.3

0.82 (0.79–0.85)

76.9

0.83 (0.80–0.86)

  EL-LR

4

74.0

0.81 (0.78–0.84)

74.3

0.83 (0.79–0.87)