Table 4 Quality of the third ICU day severity predictive ML models for eICU
From: Novel criteria to classify ARDS severity using a machine learning approach
| Algorithm | AUC, mean ± SD | CORR, mean ± SD |
|---|---|---|
| (a) PaO2/FiO2 results | ||
| Scenario I: Predicting ARDS Severity in the 3rd ICU day using the data in 1st ICU day | ||
| XGBoost | 0.712 ± 0.032 | 0.398 ± 0.061 |
| RF | 0.714 ± 0.030 | 0.393 ± 0.059 |
| LightGBM | 0.713 ± 0.028 | 0.373 ± 0.069 |
| *Scenario II: Predicting ARDS Severity in the 3rd ICU day using the data in 2nd ICU day | ||
| *XGBoost | 0.863 ± 0.016 | 0.725 ± 0.028 |
| RF | 0.863 ± 0.016 | 0.700 ± 0.040 |
| LightGBM | 0.860 ± 0.014 | 0.714 ± 0.028 |
| Scenario III: Predicting ARDS Severity in the 3rd ICU day using the data in 1st & 2nd ICU days | ||
| XGBoost | 0.860 ± 0.015 | 0.717 ± 0.025 |
| RF | 0.854 ± 0.017 | 0.693 ± 0.038 |
| LightGBM | 0.857 ± 0.014 | 0.713 ± 0.027 |
| (b) P/FPE results | ||
| Scenario I: Predicting ARDS Severity in the 3rd ICU day using the data in 1st ICU day | ||
| XGBoost | 0.735 ± 0.034 | 0.525 ± 0.056 |
| RF | 0.735 ± 0.034 | 0.514 ± 0.057 |
| LightGBM | 0.734 ± 0.034 | 0.511 ± 0.053 |
| *Scenario II: Predicting ARDS Severity in the 3rd ICU day using the data in 2nd ICU day | ||
| *XGBoost | 0.873 ± 0.022 | 0.745 ± 0.033 |
| RF | 0.868 ± 0.016 | 0.739 ± 0.039 |
| LightGBM | 0.869 ± 0.023 | 0.728 ± 0.043 |
| Scenario III: Predicting ARDS Severity in the 3rd ICU day using the data in 1st & 2nd ICU days | ||
| XGBoost | 0.872 ± 0.020 | 0.725 ± 0.040 |
| RF | 0.860 ± 0.015 | 0.731 ± 0.038 |
| LightGBM | 0.871 ± 0.022 | 0.717 ± 0.040 |