Table 1 Patient Demographics and Characteristics
From: Assessing stroke severity using electronic health record data: a machine learning approach
Characteristic | Training set (n = 6116) | Hold-out test set (n = 1033) | Overall population (n = 7149) |
|---|---|---|---|
Demographics | |||
Age, mean (SD) | 66 (14) | 67 (14) | 66 (14) |
Female, n (%) | 3196 (52) | 568 (55) | 3764 (53) |
Region | |||
Northeast, n (%) | 464 (8) | 84 (8) | 548 (8) |
Midwest, n (%) | 2388 (39) | 389 (38) | 2777 (39) |
South, n (%) | 2957 (48) | 496 (48) | 3453 (48) |
West, n (%) | 186 (3) | 40 (4) | 226 (3) |
Other/Unknown, n (%) | 121 (2) | 24 (2) | 145 (2) |
EHR data | |||
NIHSS, median (IQR) | 2 (6) | 2 (6) | 2 (6) |
LOS, median (IQR) | 3 (5) | 2 (4) | 3 (5) |
Type of strokea | |||
Ischemic, n (%) | 4328 (70.8) | 710 (68.7) | 5038 (70.5) |
Hemorrhagic, n (%) | 605 (10.0) | 113 (10.9) | 718 (10.0) |
TIA, n (%) | 2235 (36.5) | 384 (37.2) | 2619 (36.6) |
Charlson Comorbidity Indexb, median (IQR) | 1 (3) | 1 (3) | 1 (3) |