Table 1.
CNN training | Complete framework | ||
---|---|---|---|
Number | 3,445 | 379 | |
Age (years) | 57 ± 16 | 49 ± 19 | |
Sex (males) | 1,911 (55) | 228 (60) | |
Height (cm) | 176 ± 32 | 171 ± 18 | |
Weight (kg) | 79 ± 18 | 80 ± 19 | |
BMI (kg/m2) | 27 ± 5 | 27 ± 7 | |
Ethnicity | Caucasian | 2,401 (69.7) | 231 (60.9) |
Afro-Caribbean | 172 (5.0) | 54 (14.2) | |
Asian | 85 (2.5) | 10 (2.6) | |
Other | 21 (0.6) | 7 (1.8) | |
Not stated | 766 (22.2) | 77 (20.3) | |
Cardiac pathology | Healthy | 1,886 (54.7) | 68 (17.9) |
IHD | 315 (9.1) | 43 (11.3) | |
DCM | 167 (4.8) | 27 (7.1) | |
HCM | 77 (2.2) | 16 (4.2) | |
ACHD | 185 (5.4) | 59 (15.6) | |
Valvular | 133 (3.9) | 37 (9.8) | |
Vascular | 104 (3.0) | 32 (8.4) | |
Arrhythmic | 159 (4.6) | 26 (6.9) | |
Other | 419 (12.2) | 71 (18.7) |
Age, sex, height, weight, and cardiac pathology of subjects used for training of CNNs, framework and full pipeline validation. All continuous values are reported as mean ± standard deviation, while categorical variables are reported as number (percentage). ACHD, adult congenital heart disease (excluding valvular and vascular abnormalities); CNN, convolutional neural network; HCM, hypertrophic cardiomyopathy; IHD, ischaemic heart disease; SD, standard deviation.