Skip to main content

Table 3 Prediction performance of the 4 prediction models

From: Development and validation of a CT-texture analysis nomogram for preoperatively differentiating thymic epithelial tumor histologic subtypes

Training cohort AUC 95% CI Sensitivity (%) Specificity (%) Accuracy (%)
Clinical model 0.66 0.56–0.75 81.4 45.9 59.2
CT model 0.79 0.70–0.87 66.1 83.6 71.7
TA model 0.82 0.74–0.89 94.9 60.7 75.0
Combined model 0.88 0.81–0.94 93.2 67.2 79.2
Validation cohort AUC 95% CI Sensitivity (%) Specificity (%) Accuracy (%)
Clinical model 0.64 0.49–0.79 73.1 61.5 57.7
CT model 0.82 0.70–0.93 96.2 53.8 69.2
TA model 0.86 0.76–0.96 88.5 73.1 75.0
Combined model 0.94 0.89–0.98 96.2 80.8 86.5
  1. Note: AUC area under the receiver operating curve, 95% CI 95% confidence interval