Skip to main content
. 2022 Feb 3;9:100192. doi: 10.1016/j.lana.2022.100192

Table 3.

The performance comparison for Model 3 using the all the training patients and logistic regression (LR), XGBoost (XGB) and random forest (RF). All metrics were obtained by validating the models on a separate testing dataset from 40 patients. Results are shown for Models 1 and 2 separately or combined. Patient info means the information collected on the patient information document.

Elastic-net Regression-LR (%)
XGBoost - XGB (%)
Random Forest - RF (%)
ACC (SE) AUC (SE) SEN (SE) SP (SE) ACC (SE) AUC (SE) SEN (SE) SP (SE) ACC (SE) AUC (SE) SEN (SE) SP (SE)
Model 1 outputs 72 (0.070) 73 (0.083) 78 (0.098) 68 (0.099) 65 (0.075) 71 (0.086) 67 (0.111) 64 (0.102) 65 (0.075) 71 (0.084) 67 (0.111) 64 (0.102)
Model 2 outputs 92 (0.041) 95 (0.025) 94 (0.054) 91 (0.061) 90 (0.047) 92 (0.03) 89 (0.074) 91 (0.061) 90 (0.047) 92 (0.039) 89 (0.074) 91 (0.061)
Patient info 88 (0.052) 96 (0.020) 72 (0.105) 100 (0) 95 (0.034) 99 (0.006) 94 (0.054) 95 (0.044) 95 (0.034) 98 (0.01) 89 (0.074) 1 (0)
Model 1 & 2 outputs 80 (0.063) 89 (0.054) 83 (0.087) 77 (0.089) 80 (0.063) 88 (0.053) 78 (0.098) 82 (0.082) 82 (0.060) 89 (0.054) 78 (0.09) 86 (0.073)
Model 1 outputs + patient info 78 (0.066) 86 (0.061) 78 (0.09) 77 (0.089) 65 (0.075) 79 (0.071) 72 (0.106) 59 (0.104) 75 (0.068) 85 (0.065) 72 (0.105) 77 (0.089)
Model 2 outputs + patient info 90 (0.047) 96 (0.019) 89 (0.074) 91 (0.061) 88 (0.052) 93 (0.034) 83 (0.088) 91 (0.061) 88 (0.052) 96 (0.021) 83 (0.087) 91 (0.061)
All 88 (0.052) 92 (0.039) 83 (0.087) 91 (0.061) 78 (0.066) 87 (0.058) 72 (0.106) 82 (0.082) 80 (0.063) 90 (0.049) 72 (0.105) 86 (0.073)

*ACC-accuracy; AUC area under curve; SEN-sensitivity; SP-specificity; SE-standard error.

HHS Vulnerability Disclosure