Table 1 The characteristics of four datasets: MIDRC21, AREDS24, OHTS23, and MIMIC-CXR22

From: Improving model fairness in image-based computer-aided diagnosis

Disease (Dataset)

Subgroup

Attribute

Positive

Total

   

%

 

COVID-19 (MIDRC)

 

No. of images

39,369

50.55

77,887

Age

<75 yrs

34,328

52.38

65,542

> = 75 yrs

5531

44.80

12,345

Sex

Male

22,395

51.04

43,880

Female

16,974

49.91

34,007

Race

White

14,355

37.33

38,457

Black

21,292

70.20

30,239

Other races

3722

40.50

9191

Thorax abnormality (MIMIC-CXR)

 

No. of images

150,509

69.19

217,536

Age

<60 yrs

53,564

59.53

89,975

> = 60 yrs

96,945

76.00

127,561

Sex

Male

83,823

71.16

117,790

Female

66,686

66.86

99,746

Race

Other races

132,455

70.41

188,130

Black

18,054

61.40

29,406

POAG (OHTS)

 

No. of images

2327

6.22

37,399

Age

<60 yrs

420

2.58

16,254

> = 60 yrs

1907

9.04

21,085

Sex

Male

1303

8.05

16,185

Female

1024

8.71

21,154

Race

Other races

1554

5.46

28,460

Black

773

8.71

8879

Late AMD (AREDS)

 

No. of images

8521

12.90

66,060

Age

<65 yrs

276

7.31

3775

65–75 yrs

3013

9.06

33,255

> = 75 yrs

5232

18.02

29,030

Sex

Male

3768

13.16

28,623

Female

4753

12.70

37,437

Race

Other races

8496

13.31

63,808

Black

25

1.11

2252