Table 7 Comparison of diagnostic accuracy between MSI-PTDM and other models of tuberculosis
From: Clinical assistant decision-making model of tuberculosis based on electronic health records
| Â | Accuracy | AUC | Sensitivity | Specificity |
|---|---|---|---|---|
MSI-PTDM | 0.9696 (0.9657, 0.9735) | 0.9858 (0.9777, 0.9939) | 0.9318 (0.9296, 0.9340) | 0.9696 (0.9657, 0.9735) |
SS-PTDM | 0.9482 (0.9458, 0.9538) | 0.9674 (0.9356, 0.9514) | 0.8352 (0.8329, 0.8375) | 0.9483 (0.9443, 0.9523) |
US-PTDM | 0.9453 (0.9433, 0.9469) | 0.9605 (0.961, 0.9738) | 0.8284 (0.8257, 0.8311) | 0.9453 (0.9433, 0.9473) |
Text-CNN | 0.9185 (0.9128, 0.9242) | 0.9486 (0.9513, 0.9697) | 0.8251 (0.8225, 0.8277) | 0.9186 (0.9130, 0.9242) |
GRU | 0.9122 (0.9097, 0.9147) | 0.9354 (0.9414, 0.9558) | 0.8208 (0.8164, 0.8252) | 0.9122 (0.9095, 0.9149) |
LSTM | 0.9047 (0.898, 0.9114) | 0.9292 (0.9244, 0.9464) | 0.8142 (0.8081, 0.8203) | 0.9047 (0.8981, 0.9113) |
Bi_LSTM | 0.9180 (0.9133, 0.9227) | 0.9435 (0.9198, 0.9386) | 0.8104 (0.8067, 0.8141) | 0.9182 (0.9135, 0.9229) |
XGBoost | 0.9305 (0.9288, 0.9322) | 0.9571 (0.9511, 0.9631) | 0.8068 (0.8038, 0.8098) | 0.9305 (0.9288, 0.9322) |
Random Forest | 0.8996 (0.8924, 0.9068) | 0.9428 (0.9298, 0.9558) | 0.8318 (0.8250, 0.8386) | 0.8997 (0.8925, 0.9069) |
SVM | 0.9311 (0.9266, 0.9356) | 0.9429 (0.9333, 0.9525) | 0.7844 (0.7813, 0.7875) | 0.9312 (0.9267, 0.9357) |