Table 1 Analysis of sample characteristics

From: Artificial intelligence-enabled analysis of handheld single-lead electrocardiograms to predict incident atrial fibrillation: an analysis of the VITAL-AF randomized trial

 

Development set (n = 12,136)

Test set (n = 4221)*

Full inference set (n = 15,694)*

 

N(%) or Mean ± SD

Age

74.7 ± 7.1

73.8 ± 6.7

74.3 ± 6.9

Female

6,823 (56.2%)

2,538 (60.1%)

9,104 (58.0%)

Race

 Black

576 (4.7%)

245 (5.8%)

802 (5.1%)

 Hispanic

241 (2.0%)

106 (2.5%)

340 (2.2%)

 Other

946 (7.8%)

342 (8.1%)

1242 (7.9%)

 Unknown

233 (1.9%)

73 (1.7%)

297 (1.9%)

 White

10,140 (83.6%)

3,455 (81.9%)

13,013 (82.9%)

Active smoker

535 (4.4%)

183 (4.3%)

695 (4.4%)

Systolic blood pressure, mmHg

131 ± 16

131 ± 16

131 ± 16

Diastolic blood pressure, mmHg

75 ± 9

75 ± 9

75 ± 9

Height, cm

165.6 ± 10.7

164.8 ± 10.7

165.2 ± 10.7

Weight, kg

77.1 ± 18.0

75.8 ± 17.6

76.4 ± 17.7

Body mass index, kg/m2‡

28.0 ± 5.6

27.7 ± 5.4

27.9 ± 5.5

Anti-hypertensive medication use

9022 (74.3%)

2952 (69.9%)

11,380 (72.5%)

Hypertension

9539 (78.6%)

3185 (75.5%)

12,128 (77.3%)

Diabetes

3115 (25.7%)

1031 (24.4%)

3,934 (25.1%)

Myocardial infarction

1176 (9.7%)

301 (7.1%)

1393 (8.9%)

Heart failure

1891 (17.1%)

444 (10.5%)

2245 (14.3%)

CHARGE-AF score

13.7 ± 1.0

13.5 ± 0.9

13.6 ± 1.0

Prevalent AF

2133 (17.6%)

0 (0%)

0 (0%)

2-year incident AF (n, cumulative %)

309 (3.7%)

119 (3.2%)

411 (3.1%)

  1. CHARGE-AF cohorts for heart and aging research in genomic epidemiology-atrial fibrillation17
  2. *Excludes tracings read as AF or performed among individuals with prevalent AF (see text)
  3. All characteristics obtained at study enrollment
  4. Excluding missing data (see text)