Introduction
Methods
Data source and study population
Study variables
Statistical analysis
Results
Baseline characteristics
Variables | Total (n = 1094) | No ICU mortality (n = 464) | ICU mortality (n = 630) | P |
---|---|---|---|---|
Age (years) | 62.6 ± 15.4 | 62.7 ± 15.1 | 62.5 ± 15.6 | 0.874 |
Male (%) | 60.7 | 57.5 | 63.0 | 0.067 |
Mean arterial pressure (mmHg) | 81.0 ± 17.3 | 82.8 ± 16.4 | 79.7 ± 17.8 | 0.003 |
Heart rate (/min) | 105.2 ± 25.4 | 99.6 ± 24.2 | 109.3 ± 25.5 | < 0.001 |
Respiratory rate (/min) | 24.0 ± 8.4 | 21.9 ± 7.1 | 25.5 ± 8.9 | < 0.001 |
Body temperature (°C) | 36.3 ± 1.6 | 36.4 ± 1.4 | 36.3 ± 1.7 | 0.159 |
White blood cells (× 103/μL) | 14.1 ± 20.1 | 14.1 ± 11.5 | 14.1 ± 24.6 | 0.969 |
Hemoglobin (g/dL) | 9.8 ± 2.2 | 10.1 ± 2.1 | 9.6 ± 2.2 | < 0.001 |
Blood urea nitrogen (mg/dL) | 50.3 ± 29.9 | 49.0 ± 29.7 | 51.3 ± 30.0 | 0.219 |
Creatinine (mg/dL) | 2.7 ± 1.7 | 3.0 ± 2.0 | 2.5 ± 1.4 | < 0.001 |
Albumin (g/dL) | 2.8 ± 0.6 | 2.9 ± 0.6 | 2.6 ± 0.6 | < 0.001 |
pH | 7.3 ± 0.1 | 7.3 ± 0.1 | 7.3 ± 0.1 | < 0.001 |
Sodium (mEq/L) | 138.6 ± 8.1 | 138.3 ± 7.1 | 138.8 ± 8.8 | 0.318 |
Potassium (mEq/L) | 4.3 ± 0.9 | 4.2 ± 0.8 | 4.4 ± 1.0 | < 0.001 |
Target clearance (ml/min) | 42.7 ± 14.1 | 41.8 ± 13.5 | 43.3 ± 14.5 | 0.072 |
Diabetes mellitus (%) | 30.3 | 36.6 | 25.6 | < 0.001 |
Hypertension (%) | 27.8 | 29.7 | 26.3 | 0.216 |
Myocardial infarction (%) | 8.5 | 10.1 | 7.3 | 0.097 |
Chronic heart failure (%) | 15.2 | 17.7 | 13.3 | 0.048 |
Stroke (%) | 12.9 | 15.5 | 11.0 | 0.026 |
Peripheral vascular disease (%) | 7.8 | 9.1 | 6.8 | 0.174 |
Dementia (%) | 5.2 | 6.9 | 4.0 | 0.031 |
Chronic obstructive pulmonary disease (%) | 3.9 | 2.6 | 4.9 | 0.050 |
Connective tissue disease (%) | 1.4 | 1.5 | 1.3 | 0.737 |
Peptic ulcer disease (%) | 2.5 | 2.2 | 2.7 | 0.567 |
Cancer (%) | 36.8 | 34.1 | 38.9 | 0.101 |
Ischemic heart disease (%) | 12.1 | 14.7 | 10.2 | 0.024 |
Chronic kidney disease (%) | 29.7 | 41.4 | 11.4 | < 0.001 |
Ventilator apply (%) | 82.9 | 75.2 | 88.6 | < 0.001 |
Atrial fibrillation (%) | 12.3 | 16.4 | 9.4 | 0.001 |
APACHE II score | 35.9 ± 10.3 | 33.7 ± 10.4 | 37.6 ± 9.9 | < 0.001 |
SOFA score | 12.0 ± 3.6 | 10.7 ± 3.4 | 13.0 ± 3.4 | < 0.001 |
MOSAIC score | 20.8 ± 10.6 | 16.4 ± 9.2 | 24.1 ± 10.4 | < 0.001 |
Development of mortality prediction model
Models | AUC (95% CI) | P value* | P value† | P value‡ | Accuracy | F1 score |
---|---|---|---|---|---|---|
APACHE II | 0.611 (0.583–0.640) | 0.607 | 0.660 | |||
SOFA | 0.677 (0.651–0.703) | 0.629 | 0.643 | |||
MOSAIC | 0.722 (0.677–0.767) | 0.660 | 0.658 | |||
κ-Nearest neighbor | 0.762 (0.719–0.805) | < 0.001 | < 0.001 | 0.213 | 0.673 | 0.745 |
Support vector machine | 0.771 (0.729–0.813) | < 0.001 | < 0.001 | 0.119 | 0.692 | 0.752 |
Multivariate adaptive regression splines | 0.753 (0.710–0.796) | < 0.001 | 0.003 | 0.332 | 0.673 | 0.736 |
Random forest | 0.784 (0.744–0.825) | < 0.001 | < 0.001 | 0.045 | 0.690 | 0.762 |
Extreme gradient boost | 0.776 (0.735–0.818) | < 0.001 | < 0.001 | 0.085 | 0.715 | 0.763 |
Artificial neural network | 0.776 (0.735–0.818) | < 0.001 | < 0.001 | 0.082 | 0.694 | 0.749 |
Rank of predictors in the prediction model
In-hospital mortality prediction model
Models | AUC (95% CI) | P value* | P value† | P value‡ | Accuracy | F1 score |
---|---|---|---|---|---|---|
APACHE II | 0.593 (0.563–0.622) | 0.586 | 0.654 | |||
SOFA | 0.664 (0.636–0.691) | 0.603 | 0.645 | |||
MOSAIC | 0.690 (0.641–0.740) | 0.633 | 0.656 | |||
κ-Nearest neighbor | 0.721 (0.675–0.767) | < 0.001 | 0.037 | 0.379 | 0.673 | 0.776 |
Support vector machine | 0.755 (0.711–0.799) | < 0.001 | < 0.001 | 0.054 | 0.686 | 0.782 |
Multivariate adaptive regression splines | 0.756 (0.713–0.799) | < 0.001 | < 0.001 | 0.050 | 0.694 | 0.781 |
Random forest | 0.768 (0.726–0.810) | < 0.001 | < 0.001 | 0.019 | 0.700 | 0.757 |
Extreme gradient boost | 0.754 (0.709–0.798) | < 0.001 | < 0.001 | 0.062 | 0.711 | 0.790 |
Artificial neural network | 0.762 (0.719–0.806) | < 0.001 | < 0.001 | 0.032 | 0.707 | 0.790 |