Table 3 Diagnostic performance of CDA Prediction models

Regression Analysis [19]

VIllalobos-2017

2017

Expert consensus + binary logistic regression

80

N/A

87.7%

70.1%

-

-

-

-

-

-

0

Regression Analysis + AIC [31]

Vujevic-2017

2017

Logistic Regression and Akaike Information Criterion (AIC)

40

N/A

74.3%

88.3%

-

-

-

-

-

-

0

Step wise regression [20]

Ramgopal-2020 (Derivation and Validation)

2020

Regression

N/A

N/A

98.6—100%

49.2—50%

16.7—17%

99.7—100%

-

-

-

-

0.95

1

Random Forest Modelling [20]

Ramgopal-2020 (Derivation and Validation)

2020

Machine learning

N/A

N/A

99—100%

75—82%

29—36%

100%

-

-

-

-

0.96

1

Support Vector Machine Model [20]

Ramgopal-2020 (Derivation and Validation)

2020

Machine learning

N/A

N/A

97%

48—52%

16—17%

99%

-

-

-

-

0.93

1

Single-Hidden Layer Neural Network [20]

Ramgopal-2020 (Derivation and Validation)

2020

Machine learning

N/A

N/A

96—99%

69—72%

24-—26%

99%

-

-

-

-

0.95

1

Logistic Regression [32]

Chiu-2021

2021

Regression

N/A

N/A

-

-

-

-

90.0%

59.0%

-

-

0.85

0

Support Vector Machine Model [32]

Chui-2021

2021

Machine learning

N/A

N/A

-

-

-

-

91.0%

60.0%

-

-

0.84

0

Extreme Gradient Boosting [32]

Chui-2023

2021

Machine learning

N/A

N/A

-

-

-

-

90.0%

57.0%

-

-

0.84

0

Not Documented [33]

Poirier-2021

2021

Not Documented

N/A

N/A

90%

32.9%

-

98.1%

100%

30.0%

-

100%

0

Regression Analysis [22, 23, 38]

Yaeger 2021 – 2022 (Derivation and Validation)

2021 & 2022

Regression

N/A

0.01/0.03/0.05

92.9—94.6%

50—74.5%

16.1- 23.9%

99%

86.4—94.9%

47.3—51.6%

3.3- 20.5%

99%

0.92–0.95

1

Super Learner Modelling [22, 23, 38]

Yaeger 2021 – 2022 (Derivation and Validation)

2021 & 2022

Machine learning

N/A

0.01/0.03/0.06

96%

65—74%

17—23%

99.5%

91—100%

3.4—51%

2—3%

99.7—100%

0.9–0.96

1

FIRST [35]

Chong-2023

2023

AutoScore Machine Learning Model

N/A

5%/10%/15% > 

93.2% (Predicted Risk-15%)

29.9%

27.5%

94%

-

-

-

-

0.74

0

FIRST + [35]

Chong-2023

2023

AutoScore Machine Learning Model

N/A

5%/10%/15% > 

81.8% (Predicted Risk-15%)

65.6%

40.4%

92.7%

-

-

-

-

0.88

0

HRV and HRnV [34]

Chong-2023

2023

Regression

N/A

N/A

-

-

-

-

-

-

-

-

0.81

0

Deep Learning Model [21]

Yang-2023 (Derivation and Validation)

2023

Machine learning

N/A

N/A

-

-

-

-

99—100%

48—54%

5—6%

100%

0.87

1

LASSO [36]

Ballard 2024

2024

Machine Learning

N/A

-

-

-

-

-

-

-

-

0.83

0

Logistic [36]

Ballard 2024

2024

Machine Learning

N/A

-

-

-

-

-

-

-

-

0.84

0

Random Forest [36]

Ballard 2024

2024

Machine Learning

N/A

-

-

-

-

-

-

-

-

0.81

0

XGBoost [36]

Ballard 2024

2024

Machine Learning

N/A

 < 1%/ < 2%/ < 3%/ < 5%/ < 10%

-

-

-

-

-

-

-

-

0.84

0