Table 1. Patient characteristics for the most important features in the machine learning models, according to the validated dataset (n=27,561). The validated dataset represents a dataset where both the EHRsa andCTDsb agree.
| Characteristics | Values |
|---|---|
| Age (years), mean (SD) | 32 (5) |
| BMI (kg/m2), mean (SD) | 26.2 (5.3) |
| Systolic blood pressure (mm Hg), mean (SD) | 111 (11) |
| Diastolic blood pressure (mm Hg), mean (SD) | 67 (8) |
| Parity, mean (SD) | 0.9 (1.1) |
| Ethnic origin, n (%) | |
| Caucasian | 24,180 (87.8) |
| South East Asian | 1360 (4.9) |
| Black African | 554 (2.0) |
| Asian | 489 (1.8) |
| Middle Eastern | 154 (0.6) |
| Latin American | 26 (0.1) |
| Mixed | 10 (0.1) |
| Other | 788 (3) |
| Occupation skill level (ISCOc), n (%) | |
| Level 0 (unemployed) | 5254 (19) |
| Level 1 (elementary occupations) | 369 (1.3) |
| Level 2 (clerical and service) | 4404 (15.9) |
| Level 3 (technicians and associates) | 2389 (8.6) |
| Level 4 (professionals and managers) | 15,145 (55.1) |
| Family history of diabetes mellitus, n (%) | 6407 (23.3) |
| History of GDMd, n (%) | 1078 (3.9) |
| Other endocrine problems, n (%) | 5854 (21.4) |
| Prevalence of GDM, n (%) | 3188 (11.7) |
a
EHR: electronic health record.
b
CTD: clinical team database.
c
ISCO: International Standard Classification of Occupations.
d
GDM: gestational diabetes mellitus.