Fig. 5

Different models based on PET/CT group predicted ROC curves of brain metastasis of lung adenocarcinoma brain. Note: Receiver operating characteristics (ROC) curves of brain metastasis of lung adenocarcinoma predicted by PET/CT groups of different models. The support vector machine model (SVM) model performed best (AUC = 0.82), followed by the logistic regression model (LR) (AUC = 0.80), the naïve Bayes model (NB) (AUC = 0.79), the random forest model (RF) (AUC = 0.73), the neural network model (NN) (AUC = 0.67), and the adaptive boost model (AdaBoost) (AUC = 0.67). The sensitivity and specificity of the best-performing SVM models were 0.77 and 0.83, respectively