Research Article
Predicting Mortality in Hospitalized COVID-19 Patients in Zambia: An Application of Machine Learning
Table 3
Summary of categorical features with their respective proportions.
| | Categorical features | Subcategories | Total: 1,433 (100%) n (%) | Recovered: 1,304 (89.9%) n (%) | Died: 129 (10.1%) n (%) | P value |
| | Sex | Male | 911 (63.6) | 828 (63.5) | 83 (64.3) | d | | Female | 522 (36.4) | 476 (36.5) | 46 (35.7) | |
| | HIV | Positive | 276 (19.3) | 251 (19.2) | 25 (19.4) | d | | Negative | 1157 (80.7) | 1053 (80.8) | 104 (80.6) | |
| | Diabetes | Yes | 352 (24.6) | 303 (23.2) | 49 (38.0) | d | | No | 1081 (75.4) | 1001 (76.8) | 80 (62.0) | |
| | Hypertension | Yes | 729 (50.9) | 644 (49.4) | 85 (65.9) | d | | No | 704 (49.1) | 660 (50.6) | 44 (34.1) | |
| | Wave | 1st-wave | 534 (37.3) | 506 (38.8) | 28 (21.7) | d | | 2nd-wave | 566 (39.5) | 505 (38.7) | 61 (47.3) | | | 3rd-wave | 333 (23.2) | 293 (22.5) | 40 (31.0) | |
| | Ward | General | 496 (34.6) | 453 (34.7) | 43 (33.3) | d | | Suspect | 250 (17.4) | 242 (18.6 | 8 (6.2) | | | Infectious dis. | 251 (17.5) | 216 (16.6 | 35 (27.1) | | | High cost | 293 (20.4) | 277 (21.2 | 16 (12.4) | | | ICU | 143 (10.0) | 116 (8.9) | 27 (20.9) | |
| | Smoking | Yes | 65 (4.5) | 64 (4.9) | 1 (0.8) | d | | No | 1368 (95.5) | 1240 (95.1) | 128 (99.2) | |
| | TB | Yes | 68 (4.7) | 64 (4.9) | 4 (3.1) | d | | No | 1365 (95.3) | 1240 (95.1) | 125 (96.9) | |
| | CKD | Yes | 32 (2.2) | 28 (2.1) | 4 (3.1) | d | | No | 1401 (97.8) | 1276 (97.9) | 125 (96.9) | |
| | Alcohol | Yes | 204 (14.2) | 185 (14.2) | 19 (14.7) | d | | No | 1229 (85.8) | 1119 (85.8) | 110 (85.3) | |
| | CPD | Yes | 21 (1.5) | 13 (1.0) | 8 (6.2) | d | | No | 1412 (98.5) | 1291 (99.0) | 1296 (93.8) | |
|
|
Note: d value from Mann–Whitney U test; features omitted in models with selected features. |
