Table 3 Comparison of the prediction performances of the prediction models on the test dataset.
From: Artificial intelligence to predict in-hospital mortality using novel anatomical injury score
Model | TN | FP | FN | TP | Sensitivity (%) | Specificity (%) | Accuracy (%) | Balanced accuracy (%) | AUROC |
|---|---|---|---|---|---|---|---|---|---|
LR (AIS) | 3200 | 449 | 32 | 95 | 74.80 | 87.70 | 87.26 | 81.25 | 0.8770 |
RF (AIS) | 2720 | 929 | 20 | 107 | 84.25 | 74.54 | 74.87 | 79.40 | 0.8598 |
SVM (AIS) | 3032 | 617 | 21 | 106 | 83.46 | 83.01 | 83.10 | 83.28 | 0.8943 |
DNN (AIS) | 3230 | 419 | 33 | 94 | 74.02 | 88.52 | 88.03 | 81.27 | 0.8819 |
LR (Region-6) | 3059 | 590 | 25 | 102 | 80.32 | 83.83 | 83.72 | 82.07 | 0.8819 |
RF (Region-6) | 3090 | 559 | 24 | 103 | 81.10 | 84.68 | 84.56 | 82.89 | 0.8867 |
SVM (Region-6) | 3009 | 640 | 23 | 104 | 81.89 | 82.46 | 82.44 | 82.18 | 0.8712 |
DNN (Region-6) | 3028 | 621 | 20 | 107 | 84.25 | 82.98 | 83.02 | 83.62 | 0.8871 |
LR (Region-46) | 3109 | 540 | 24 | 103 | 81.10 | 85.20 | 85.06 | 83.15 | 0.9013 |
RF (Region-46) | 3054 | 595 | 23 | 104 | 81.89 | 83.69 | 83.63 | 82.79 | 0.8853 |
SVM (Region-46) | 3091 | 558 | 23 | 104 | 81.89 | 84.71 | 84.61 | 83.30 | 0.8829 |
DNN (Region-46) | 3161 | 488 | 21 | 106 | 83.46 | 86.63 | 86.52 | 85.05 | 0.9084 |
ISS-16 | 2944 | 705 | 25 | 102 | 80.31 | 80.68 | 80.67 | 80.50 | 0.8709 |
ISS-25 | 3387 | 262 | 65 | 62 | 48.82 | 92.82 | 91.34 | 70.82 | |
NISS-16 | 2618 | 1031 | 17 | 110 | 86.61 | 71.75 | 72.25 | 79.18 | 0.8681 |
NISS-25 | 3241 | 408 | 44 | 83 | 65.35 | 88.82 | 88.03 | 77.09 |
