Skip to main content

Table 2 Sample characteristics of the US population-based sample (n = 1143) and the Swiss clinic-based sample (n = 2481)

From: Machine learning techniques for personalized breast cancer risk prediction: comparison with the BCRAT and BOADICEA models

Variables included in BCRAT and BOADICEA models and in ML algorithmsUS population-based sample n = 1143Swiss clinic-based sample n = 2481
Age (range)50.86 ± 6.22 (35–64)50.78 ± 12.77 (13–89)
Age at menarche (range)12.56 ± 1.54 (8–18)12.91 ± 1.59 (8–18)
Age at first live birth (range)24.29 ± 5.62 (13–42)24.13 ± 5.72 (15–48)
Number of biopsies (n = 847)1.20 ± 1.21
 Atypical hyperplasia14 (1.65%)
Breast cancer850 (74.37%)886 (35.71%)
 First-ductal carcinoma in situ (DCIS)434 (51.06%)50 (5.64%)
 First-invasive breast cancer404 (47.52%)807 (91.08%)
 First-breast cancer age onset (range)40.03 ± 4.79 (26–54)46.07 ± 10.69 (22–84)
 Bilateral breast cancer4 (0.47%)160 (18.06%)
 Estrogen receptor (ER) positive618 (69.75%)
 Progesterone receptor (PR) positive561 (63.32%)
Pancreatic cancer13 (0.52%)
 Pancreatic cancer age onset (range) 55.10 ± 9.35 (36–75)
Ovarian cancer9 (0.79%)133 (5.36%)
 Ovarian cancer age onset (range)45.83 ± 5.00 (36–50)56.44 ± 13.16 (21–85)
 Having also breast cancer420
Ethnicity (% Black)401 (35.08%)71 (2.86%)
Ashkenazi Jewish origin12 (1.05%)65 (2.29%)
Number of first-degree relatives with breast cancer0.98 ± 1.050.25 ± 0.55
 Breast cancer patients0.81 ± 1.05
 Relatives of breast cancer patients1.49 ± 0.88
BRCA1 or BRCA2 germline mutations32 (2.79%) 235 tested209 (8.42%) 1052 tested
  1. – Data not available