Table 3 Feature importance rank for the machine learning models. The mean contribution for each feature as percentages is shown in parentheses. The combined contribution of the top 5, 10, and 15 feature are shown in bold face
Rank | XGB | ADA | LR |
|---|---|---|---|
1 | Age (0.096) | Age (0.106) | Age (0.096) |
2 | Vent (0.089) | Vent (0.077) | Source\(^b\) (0.084) |
3 | Verbal (0.077) | HR (0.068) | Vent (0.075) |
4 | BUN (0.068) | BUN (0.057) | Adx (0.067) |
5 | Motor (0.058) | Elective\(^a\) (0.051) | HR (0.061) |
Top 5 | 0.390 | 0.360 | 0.382 |
6 | HR (0.055) | Temp (0.051) | Verbal (0.057) |
7 | RR (0.049) | Verbal (0.050) | RR (0.053) |
8 | WBC (0.041) | WBC (0.049) | Eyes (0.051) |
9 | Temp (0.040) | RR (0.049) | BUN (0.045) |
10 | Adx (0.038) | Source\(^b\) (0.043) | Op/Non-Op (0.040) |
Top 10 | 0.613 | 0.602 | 0.628 |
11 | Elective\(^a\) (0.034) | Motor (0.036) | Motor (0.039) |
12 | Mean BP (0.028) | Adx (0.034) | Albumin (0.038) |
13 | Source\(^b\) (0.028) | FiO2 (0.29) | WBC (0.030) |
14 | Creatinine (0.026) | Mean BP (0.029) | Intubated (0.029) |
15 | Eyes (0.026) | Sodium (0.028) | Temp (0.027) |
Top 15 | 0.754 | 0.758 | 0.791 |