Table 4.
Performance summary of the machine learning models for VRE, CRE, MRSA, and MDRO colonization predictions.
| Threshold opt. upper bound = 0.05 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.010 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.03 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.1 | MDRO (792/4503 = 17.59%) | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.77 | Training AUC | 0.70 | 0.76 | 0.78 | Training AUC | 0.65 | 0.66 | 0.66 | Training AUC | 0.72 | 0.86 | 0.88 | 0.75 |
| Testing AUC | 0.80 | 0.77 | 0.77 | Testing AUC | 0.78 | 0.72 | 0.71 | Testing AUC | 0.66 | 0.66 | 0.69 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.76 |
| Testing sensitivity | 0.99 | 0.97 | 1.00 | Testing sensitivity | 1.00 | 0.73 | 0.82 | Testing sensitivity | 1.00 | 0.88 | 1.00 | Testing sensitivity | 0.92 | 0.97 | 1.00 | 0.93 |
| Testing specificity | 0.09 | 0.33 | 0.15 | Testing specificity | 0.31 | 0.68 | 0.37 | Testing specificity | 0.02 | 0.22 | 0.00 | Testing specificity | 0.30 | 0.39 | 0.21 | 0.51 |
| Threshold opt. upper bound = 0.1 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.015 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.075 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.2 | MDRO (792/4503 = 17.59%) | |||||||||
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.76 | Training AUC | 0.70 | 0.76 | 0.80 | Training AUC | 0.65 | 0.66 | 0.67 | Training AUC | 0.72 | 0.86 | 0.88 | 0.76 |
| Testing AUC | 0.80 | 0.77 | 0.77 | Testing AUC | 0.78 | 0.72 | 0.71 | Testing AUC | 0.66 | 0.66 | 0.71 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.81 |
| Testing sensitivity | 0.89 | 0.79 | 0.88 | Testing sensitivity | 0.82 | 0.73 | 0.73 | Testing sensitivity | 0.76 | 0.71 | 0.82 | Testing sensitivity | 0.68 | 0.82 | 0.90 | 0.93 |
| Testing specificity | 0.49 | 0.57 | 0.48 | Testing specificity | 0.57 | 0.77 | 0.51 | Testing specificity | 0.45 | 0.53 | 0.45 | Testing specificity | 0.59 | 0.74 | 0.69 | 0.58 |
| Threshold opt. upper bound = 0.15 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.020 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.1 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.3 | MDRO (792/4503 = 17.59%) | |||||||||
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.76 | Training AUC | 0.70 | 0.76 | 0.80 | Training AUC | 0.65 | 0.66 | 0.66 | Training AUC | 0.72 | 0.86 | 0.87 | 0.76 |
| Testing AUC | 0.80 | 0.77 | 0.78 | Testing AUC | 0.78 | 0.72 | 0.73 | Testing AUC | 0.66 | 0.66 | 0.68 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.81 |
| Testing sensitivity | 0.80 | 0.73 | 0.78 | Testing sensitivity | 0.73 | 0.55 | 0.73 | Testing sensitivity | 0.64 | 0.67 | 0.73 | Testing sensitivity | 0.65 | 0.75 | 0.85 | 0.89 |
| Testing specificity | 0.66 | 0.65 | 0.59 | Testing specificity | 0.69 | 0.83 | 0.63 | Testing specificity | 0.60 | 0.60 | 0.57 | Testing specificity | 0.63 | 0.82 | 0.79 | 0.65 |
| Threshold opt. upper bound = 0.2 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.025 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.15 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.4 | MDRO (792/4503 = 17.59%) | |||||||||
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.77 | Training AUC | 0.70 | 0.76 | 0.79 | Training AUC | 0.65 | 0.66 | 0.67 | Training AUC | 0.72 | 0.86 | 0.87 | 0.76 |
| Testing AUC | 0.80 | 0.77 | 0.77 | Testing AUC | 0.78 | 0.72 | 0.71 | Testing AUC | 0.66 | 0.66 | 0.69 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.79 |
| Testing sensitivity | 0.66 | 0.63 | 0.75 | Testing sensitivity | 0.73 | 0.36 | 0.64 | Testing sensitivity | 0.48 | 0.56 | 0.71 | Testing sensitivity | 0.57 | 0.69 | 0.82 | 0.79 |
| Testing specificity | 0.76 | 0.75 | 0.66 | Testing specificity | 0.73 | 0.89 | 0.67 | Testing specificity | 0.75 | 0.71 | 0.58 | Testing specificity | 0.73 | 0.86 | 0.83 | 0.64 |
| Threshold opt. upper bound = 0.3 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.030 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.2 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.5 | MDRO (792/4503 = 17.59%) | |||||||||
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.77 | Training AUC | 0.70 | 0.76 | 0.79 | Training AUC | 0.65 | 0.66 | 0.66 | Training AUC | 0.72 | 0.86 | 0.88 | 0.76 |
| Testing AUC | 0.80 | 0.77 | 0.77 | Testing AUC | 0.78 | 0.72 | 0.71 | Testing AUC | 0.66 | 0.66 | 0.69 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.80 |
| Testing sensitivity | 0.66 | 0.65 | 0.61 | Testing sensitivity | 0.64 | 0.36 | 0.64 | Testing sensitivity | 0.70 | 0.45 | 0.67 | Testing sensitivity | 0.56 | 0.70 | 0.85 | 0.77 |
| Testing specificity | 0.78 | 0.72 | 0.74 | Testing specificity | 0.78 | 0.91 | 0.74 | Testing specificity | 0.55 | 0.78 | 0.59 | Testing specificity | 0.75 | 0.84 | 0.79 | 0.70 |
| Threshold opt. upper bound = 0.5 | VRE (503/3860 = 13.03%) | Threshold opt. upper bound = 0.050 | CRE (53/3661 = 1.45%) | Threshold opt. upper bound = 0.3 | MRSA (332/4446 = 7.47%) | Threshold opt. upper bound = 0.6 | MDRO (792/4503 = 17.59%) | |||||||||
| Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Log. | XGBoost | Rand. | Dec. | ||||
| Reg. | Forest | Reg. | Forest | Reg. | Forest | Reg. | Forest | Tree | ||||||||
| Training AUC | 0.76 | 0.77 | 0.77 | Training AUC | 0.70 | 0.76 | 0.79 | Training AUC | 0.65 | 0.65 | 0.66 | Training AUC | 0.72 | 0.86 | 0.88 | 0.77 |
| Testing AUC | 0.80 | 0.77 | 0.78 | Testing AUC | 0.78 | 0.72 | 0.72 | Testing AUC | 0.66 | 0.66 | 0.70 | Testing AUC | 0.70 | 0.87 | 0.89 | 0.80 |
| Testing sensitivity | 0.61 | 0.59 | 0.63 | Testing sensitivity | 0.55 | 0.27 | 0.64 | Testing sensitivity | 0.48 | 0.24 | 0.76 | Testing sensitivity | 0.58 | 0.70 | 0.84 | 0.89 |
| Testing specificity | 0.82 | 0.77 | 0.74 | Testing specificity | 0.82 | 0.94 | 0.79 | Testing specificity | 0.75 | 0.89 | 0.59 | Testing specificity | 0.72 | 0.85 | 0.79 | 0.62 |