Table 2 The sensitivity and specificity of machine learning models in papillary thyroid carcinoma (PTC) prediction applied to different subgroups and different variables used for model construction

LNM

Model

 LR(Logistic regression)

19,597

55,631

0.71(0.69–0.73)

0.77(0.74–0.79)

5173

12,845

0.72(0.70–0.74)

0.77(0.73–0.80)

LASSO

357

1113

0.66–0.73

0.60–0.70

177

588

0.69–0.74

0.55–0.73

RF(Random Forest)

2536

5082

0.76(0.54–0.90)

0.59(0.53–0.65)

215

414

0.72–0.78

0.67–0.68

GBM(Gradient Boosting Machine)

2133

4601

0.49(0.30–0.69)

0.72(0.64–0.78)

32

118

0.87

0.86

 SVM(Support Vector Machine)

1659

2765

0.81–0.89

0.53–0.60

215

414

0.80–0.87

0.56–0.58

ANN(Artificial Neural Network)

2311

4036

0.69–0.82

0.44–0.66

215

414

0.79–0.81

0.51–0.57

 KNN(K-Nearest Neighbor)

1360

2861

0.32–0.90

0.36–0.63

DT(Decision Tree)

2311

4036

0.60–0.93

0.26–0.62

215

414

0.94–0.96

0.23–0.24

BN(Batch Normalization)

524

950

0.79

0.61

215

414

0.61–0.88

0.48–0.73

 DL(Deep Learning)

225

1046

0.44

0.74

Overall

33,013

82,121

0.71(0.69–0.74)

0.74(0.72–0.76)

6457

15,621

0.74(0.72–0.76)

0.74(0.70–0.77)

LR Variable

Clinical characteristics

17,648

49,435

0.71(0.69–0.73)

0.76(0.74–0.78)

4270

10,032

0.71(0.68–0.73)

0.75(0.72–0.78)

Radiomics

308

591

0.71–0.79

0.75–0.86

145

359

0.75(0.66–0.81)

0.84(0.78–0.88)

Clinical characteristics + Genomics

1137

3768

0.69(0.59–0.76)

0.76(0.41–0.93)

379

1070

0.73–0.90

0.63–0.96

Rdiomics + Ultrasonics

55

600

0.87

0.93

31

286

0.81

0.95

Ultrasonics + Clinical characteristics

131

300

0.82–0.84

0.82–0.94

208

420

0.78(0.71–0.83)

0.89(0.86–0.92)

Rdiomics + Clinical characteristics

318

637

0.65(0.59–0.71)

0.77(0.71–0.83)

140

255

0.69(0.63–0.75)

0.54(0.44–0.63)

Central LNM

Model

LR(Logistic regression)

17,648

49,165

0.69(0.66–0.71)

0.77(0.74–0.79)

4869

11,217

0.68(0.66–0.71)

0.76(0.74–0.78)

LASSO

160

439

0.71

0.7

77

220

0.74

0.73

RF(Random Forest)

1401

3267

0.39–0.88

0.60–0.63

215

414

0.72–0.78

0.67–0.68

GBM(Gradient Boosting Machine)

998

2786

0.46(0.26–0.66)

0.70(0.62–0.76)

32

118

0.87

0.86

SVM(Support Vector Machine)

524

950

0.89

0.83

215

414

0.80–0.87

0.56–0.58

ANN(Artificial Neural Network)

1176

2221

0.69–0.82

0.54–0.66

215

414

0.79–0.81

0.51–0.57

KNN(K-Nearest Neighbor)

225

1046

0.32

0.63

DT(Decision Tree)

1176

2221

0.60–0.93

0.26–0.62

215

414

0.94–0.96

0.23–0.24

BN(Batch Normalization)

524

950

0.79

0.61

215

414

0.61–0.88

0.48–0.73

 DL(Deep Learning)

225

1046

0.44

0.74

Overall

24,057

64,091

0.68(0.65–0.71)

0.74(0.72–0.76)

6053

13,625

0.74(0.71–0.76)

0.72(0.68–0.76)

LR Variable

Clinical characteristics

14,775

39,017

0.68(0.66–0.71)

0.76(0.74–0.78)

4110

9340

0.70(0.67–0.73)

0.76(0.72–0.79)

Radiomics

218

311

0.71–0.77

0.75–0.86

81

132

0.70–0.73

0.79–0.88

Clinical characteristics + Genomics

960

2570

0.66–0.82

0.67–0.97

379

1070

0.73–0.90

0.63–0.96

Rdiomics + Clinical characteristics

318

637

0.65(0.59–0.71)

0.77(0.71–0.83)

140

255

0.69(0.63–0.75)

0.54(0.44–0.63)

Ultrasonics + Clinical characteristics

104

300

0.84

0.82

112

420

0.71–0.83

0.82–0.88

Lateral Cervical LNM

Model

LR(Logistic regression)

3222

12,796

0.78(0.75–0.81)

0.79(0.70–0.85)

304

1628

0.77(0.72–0.82)

0.84(0.77–0.89)

LASSO

197

674

0.66–0.73

0.60–0.66

100

368

0.69

0.55

RF(Random Forest)

1135

1815

0.89

0.49

GBM(Gradient Boosting Machine)

1135

1815

0.55

0.77

SVM(Support Vector Machine)

1135

1815

0.81

0.6

ANN(Artificial Neural Network)

1135

1815

0.79

0.44

KNN(K-Nearest Neighbor)

1135

1815

0.9

0.36

DT(Decision Tree)

1135

1815

0.84

0.41

Overall

10,229

24,360

0.78(0.75–0.81)

0.74(0.66–0.81)

404

1996

0.77(0.72–0.82)

0.82(0.74–0.88)

LR Variable

Clinical characteristics

2873

10,418

0.78(0.75–0.81)

0.78(0.69–0.85)

160

695

0.76(0.69–0.82)

0.74(0.66–0.82)

Radiomics

90

280

0.79

0.81

64

227

0.74–0.88

0.80–0.88

Clinical characteristics + Genomics

177

1198

0.61

0.3

Rdiomics + Ultrasonics

55

600

0.87

0.93

31

286

80.6

94.5

Ultrasonics + Clinical characteristics

27

300

0.87

0.93

49

420

0.73–0.87

0.89–0.93