Research Article
Deep-Learning-Based Survival Prediction of Patients in Coronary Care Units
Table 1
Baseline demographic and laboratory characteristics of patients.
| Variable | Training cohort | Testing cohort | | () | () |
| Disease (%) | | | 0.742 | Myocardial infarction | 1477 (62.3) | 629 (61.8) | Heart failure | 498 (21.0) | 226 (22.2) | Tachycardia | 164 (6.9) | 67 (6.6) | Respiratory failure | 123 (5.2) | 44 (4.3) | Valve disorder | 109 (4.6) | 51 (5.0) | Race (%) | | | 0.321 | White | 1659 (70.0) | 698 (68.6) | Black | 136 (5.7) | 64 (6.3) | Asia | 27 (1.1) | 13 (1.3) | Hispanic | 34 (1.4) | 25 (2.5) | Others | 50 (2.1) | 26 (2.6) | Unknown | 465 (19.6) | 191 (18.8) | Anion gap (mEq/L, %) | | | 0.104 | 8~16 | 1092 (46.1) | 464 (45.6) | <8 | 130 (5.5) | 39 (3.8) | >16 | 1149 (48.5) | 514 (50.5) | Glucose (mg/dL, %) | | | 0.377 | 70~140 | 1481 (62.5) | 658 (64.7) | <70 | 45 (1.9) | 15 (1.5) | >140 | 845 (35.6) | 344 (33.8) | Neutrophil (%) | | | 0.868 | 50~70 | 366 (15.4) | 163 (16.0) | <50 | 50 (2.1) | 23 (2.3) | >70 | 1955 (82.5) | 831 (81.7) | White blood cell (K/uL, %) | | | 0.611 | 4~10 | 1947 (82.1) | 838 (82.4) | <4 | 5 (0.2) | 4 (0.4) | >10 | 419 (17.7) | 175 (17.2) | Potassium (mEq/L, %) | | | 0.659 | 3.5~5.5 | 1933 (81.5) | 841 (82.7) | <3.5 | 4 (0.2) | 1 (0.1) | >5.5 | 434 (18.3) | 175 (17.2) | Creatine kinase (%) | | | 0.013 | 18~198 | 1126 (47.5) | 529 (52.0) | <18 | 17 (0.7) | 2 (0.2) | >198 | 1228 (51.8) | 486 (47.8) | Blood urea nitrogen (mg/dL, %) | | | 0.869 | 6~22 | 764 (32.2) | 320 (31.5) | <6 | 11 (0.5) | 4 (0.4) | >22 | 1596 (67.3) | 693 (68.1) | (%) | 1479/892 (62.4/37.6) | 609/408 (59.9/40.1) | 0.183 | Age (median [IQR]) | 70 [59,79] | 70 [59,79] | 0.769 | APS-III score (mean (SD)) | 42.80 (19.38) | 42.12 (18.94) | 0.348 | SOFA score (mean (SD)) | 3.76 (2.93) | 3.76 (2.96) | 0.976 | (%) | 1311/1060 (55.3/44.7) | 568/449 (55.9/44.1) | 0.794 |
|
|