Table 2 Model performances for predicting post-filter ionized calcium levels
Labels | Models | Precision (%) | Recall (%) | F1-score (%) | Accuracy (%) |
|---|---|---|---|---|---|
“0”: < 0.25 mmol/L | AdaBoost | 70.43 | 69.61 | 69.99 | 77.94 |
XGBoost | 83.73 | 79.41 | 81.20 | 86.76 | |
SVM | 91.13 | 67.65 | 71.22 | 83.82 | |
Shallow neural network | 90.76 | 90.77 | 90.77 | 90.76 | |
“1”: 0.25–0.35 mmol/L | AdaBoost | 67.39 | 59.21 | 60.04 | 76.32 |
XGBoost | 82.65 | 78.54 | 86.41 | 85.17 | |
SVM | 92.93 | 77.41 | 79.72 | 86.25 | |
Shallow neural network | 88.45 | 88.4 | 88.40 | 88.45 | |
“2”: 0.35–0.5 mmol/L | AdaBoost | 70.22 | 59.11 | 59.85 | 77.17 |
XGBoost | 83.77 | 80.92 | 82.17 | 83.77 | |
SVM | 81.07 | 80.89 | 81.87 | 81.89 | |
Shallow neural network | 83.74 | 80.92 | 82.17 | 88.77 | |
“3”: > 0.5 mmol/L | AdaBoost | 73.15 | 68.40 | 70.03 | 79.69 |
XGBoost | 83.91 | 81.94 | 82.85 | 87.50 | |
SVM | 91.38 | 68.75 | 72.56 | 84.38 | |
Shallow neural network | 88.98 | 88.96 | 88.96 | 88.96 |