Skip to main content

Table 1 Characteristics of ten-year and 2015-year cohorts

From: Multi-vendor evaluation of artificial intelligence as an independent reader for double reading in breast cancer screening on 275,900 mammograms

Characteristics

Ten-year cohort (2009–2019)

2015-year cohort (2015)

Number of cases

Proportion of study population

Number of cases

Proportion of study population

Totala

275,900

100.0%

45,675

100.0%

Center / Vendor

MK / IMS Giotto

83,410

30.2%

10,462

22.9%

NUH / GE

69,045

25.0%

10,983

24.0%

LTHT / Hologic

64,645

23.4%

10,717

23.5%

ULH / Siemens

58,800

21.3%

13,513

29.6%

Age

 < 40

483

0.2%

5

 < 0.1%

40—49

37,696

13.7%

5,575

12.2%

50—59

114,524

41.5%

19,399

42.5%

60—69

98,289

35.6%

16,772

36.7%

70—79

23,359

8.5%

3,702

8.1%

80—89

1,534

0.6%

221

0.5%

 > 90

15

 < 0.1%

1

 < 0.1%

Positives

Total positivesb

2,683

0.97%

457

1.00%

 Screen-detected positives

2,310

0.84%

365

0.80%

 Interval cancers (ICs)

373

0.14%

92

0.20%

  Three-year ICs from UK

289

0.10%

80

0.18%

  Two-year ICs from HU

84

0.03%

12

0.03%

  1. Total number of cases for which CDR, recall rate, and arbitration rate were calculated on
  2. b Used for sensitivity, CDR, and PPV calculations