Skip to main content

Table 3 Calibration of ANN and CLR in modeling and testing datasets with 16- and 6-variable models

From: Hip fracture risk assessment: artificial neural network outperforms conditional logistic regression in an age- and sex-matched case control study

 

1

2

3

4

5

6

7

8

9

10

Mean

SD

p

Modeling

             

Chi-square

             

ANN 16v

5.791

10.735

6.784

12.737

5.859

4.315

6.067

9.698

5.161

6.981

7.413

2.583

0.013*

CLR 16v

18.458

19.948

9.761

9.008

9.222

12.553

10.714

18.406

10.518

17.758

13.635

4.221

ANN 6v

9.077

6.323

6.482

9.398

7.679

3.729

6.560

10.786

8.217

6.973

7.522

1.877

0.333

CLR 6v

14.913

5.333

5.125

6.997

12.961

4.859

11.817

9.667

12.676

3.386

8.773

3.914

ICC

             

ANN 16v

0.994

0.989

0.991

0.984

0.992

0.994

0.992

0.99

0.993

0.995

0.991

0.003

0.066

CLR 16v

0.984

0.977

0.992

0.993

0.992

0.987

0.991

0.981

0.991

0.979

0.987

0.006

ANN 6v

0.992

0.995

0.992

0.995

0.994

0.996

0.996

0.993

0.993

0.994

0.994

0.001

0.066

CLR 6v

0.986

0.992

0.996

0.995

0.985

0.994

0.989

0.98

0.988

0.998

0.990

0.005

Testing

             

Chi-square

             

ANN 16v

9.365

4.227

7.363

6.317

6.281

5.044

9.150

2.706

7.778

6.576

6.481

1.984

0.047*

CLR 16v

8.618

6.884

15.622

8.691

9.798

6.046

4.914

7.248

8.566

12.228

8.862

2.968

ANN 6v

7.334

7.647

8.936

2.493

8.714

15.144

14.947

2.043

8.193

3.678

7.913

4.309

0.646

CLR 6v

8.828

8.713

12.350

10.182

10.132

7.305

9.845

4.358

5.295

6.889

8.390

2.315

ICC

             

ANN 16v

0.927

0.985

0.976

0.979

0.975

0.944

0.952

0.995

0.942

0.966

0.964

0.021

0.007*

CLR 16v

0.906

0.944

0.912

0.949

0.952

0.883

0.965

0.965

0.907

0.922

0.931

0.027

ANN 6v

0.957

0.966

0.965

0.976

0.961

0.883

0.874

0.996

0.942

0.983

0.950

0.039

0.037*

CLR 6v

0.926

0.915

0.936

0.929

0.933

0.888

0.893

0.979

0.955

0.961

0.932

0.027

  1. * Statistically significant difference.