Table 2 Diagnostic evaluation indices and the evaluation of consistency with pathological diagnosis in the deep learning group and endoscopist group before and after propensity score matching
CAG versus CNAG | Before matching (338 vs. 793) | After matching (338 vs. 338) | ||
|---|---|---|---|---|
DL | Endoscopist | DL | Endoscopist | |
Sensitivity | 84.02% | 62.72% | 84.02% | 62.72% |
Specificity | 96.34% | 80.45% | 97.04% | 81.95% |
PV+ | 90.73% | 57.77% | 96.60% | 77.66% |
PV− | 93.40% | 83.51% | 85.86% | 68.73% |
Accuracy | 92.66% | 75.15% | 90.53% | 72.34% |
Youden index | 80.36% | 43.17% | 81.06% | 44.67% |
Odd product | 91.71 | 6.93 | 172.5 | 7.64 |
LR + | 22.96 | 3.21 | 28.39 | 3.47 |
LR− | 0.17 | 0.46 | 0.16 | 0.45 |
AUC (95% CI) | 0.906 (0.882–0.930) | 0.735 (0.700–0.769) | 0.909 (0.884–0.934) | 0.740 (0.702–0.778) |
Kappa | 0.842 | 0.492 | 0.852 | 0.558 |