Table 3 Performance summary of different models for prediction of FTC.
Model | Accuracy | AUC | Sensitivity | Specificity | F1 | PPV | NPV |
|---|---|---|---|---|---|---|---|
Test cohort | |||||||
MobileNetV2 | 0.63 (0.51–0.74) | 0.69 (0.57–0.8) | 0.71 (0.6–0.82) | 0.56 (0.39–0.75) | 0.63 (0.51–0.74) | 0.56 (0.39–0.71) | 0.71 (0.52–0.87) |
ResNet101 | 0.62 (0.51–0.74) | 0.64 (0.52–0.75) | 0.61 (0.49–0.73) | 0.64 (0.49–0.79) | 0.59 (0.47–0.71) | 0.57 (0.39–0.75) | 0.68 (0.51–0.83) |
VGG16 | 0.63 (0.51–0.74) | 0.74 (0.63–0.85) | 0.89 (0.82–0.97) | 0.41 (0.26–0.58) | 0.68 (0.56–0.79) | 0.54 (0.40–0.69) | 0.83 (0.63-1.00) |
ResNet152 | 0.67 (0.56–0.79) | 0.77 (0.67–0.87) | 0.75 (0.64–0.86) | 0.61 (0.44–0.78) | 0.67 (0.55–0.78) | 0.6 (0.43–0.76) | 0.76 (0.59–0.92) |
ResNet50 | 0.63 (0.51–0.74) | 0.69 (0.57–0.8) | 0.75 (0.64–0.86) | 0.53 (0.37–0.68) | 0.64 (0.52–0.75) | 0.55 (0.39–0.71) | 0.73 (0.57–0.88) |
Train cohort | |||||||
MobileNetV2 | 0.78 (0.73–0.83) | 0.88 (0.84–0.92) | 0.8 (0.75–0.85) | 0.77 (0.70–0.83) | 0.77 (0.71–0.82) | 0.73 (0.65–0.81) | 0.83 (0.77–0.89) |
ResNet101 | 0.84 (0.79–0.88) | 0.91 (0.87–0.94) | 0.85 (0.81–0.9) | 0.82 (0.76–0.89) | 0.82 (0.78–0.87) | 0.79 (0.72-86) | 0.88 (0.82–0.93) |
VGG16 | 0.66 (0.6–0.72) | 0.72 (0.66–0.77) | 0.65 (0.59–0.7) | 0.67 (0.59–0.74) | 0.63 (0.57–0.68) | 0.6 (0.51–0.69) | 0.71 (0.63–0.78) |
ResNet152 | 0.84 (0.8–0.89) | 0.93 (0.9–0.96) | 0.87 (0.83–0.91) | 0.82 (0.76–0.87) | 0.83 (0.79–0.88) | 0.8 (0.73–0.86) | 0.89 (0.84–0.94) |
ResNet50 | 0.87 (0.83–0.91) | 0.95 (0.92–0.97) | 0.88 (0.84–0.92) | 0.86 (0.80–0.92) | 0.86 (0.81–0.9) | 0.84 (0.77–0.90) | 0.9 (0.85–0.95) |
