Skip to main content

Table 3 Discrimination performance of radiomic features built by using the SVM, Decision tree, Random forest, X-Gradient boost and multivariable Logistic regression for the training and validation cohorts

From: CT radiomic features for predicting resectability of oesophageal squamous cell carcinoma as given by feature analysis: a case control study

Model Discrimination
Accuracy AUC ± SD F-1score
The training cohort
 SVM 0.80 0.86 ± 0.03 0.81
 Decision tree 0.69 0.73 ± 0.06 0.71
 Random forest 0.73 0.80 ± 0.07 0.75
 X-Gradient boost 0.78 0.87 ± 0.06 0.81
 MLR 0.87 0.92 ± 0.04 0.93
The validation cohort
 SVM 0.79 0.82 ± 0.03 0.80
 Decision tree 0.69 0.66 ± 0.03 0.70
 Random forest 0.67 0.67 ± 0.03 0.68
 X-Gradient boost 0.79 0.84 ± 0.03 0.79
 MLR 0.86 0.87 ± 0.02 0.86
  1. SD standard deviation, SVM support vector machine, MLR multivariable logistic regression, AUC receiver operating characteristic curve