Model Type | AUC (Training Set) | AUC (Validation Set 1) | AUC (Validation Set 2) | p-value |
|---|
Traditional Radiomics | 0.72 | 0.65 | 0.66 | 0.317 |
Convolutional Neural Network | 0.78 | 0.71 | 0.65 | 0.049 |
Support Vector Machine | 0.75 | 0.68 | 0.63 | 0.049 |
Our Integrated Model | 0.85 | 0.73 | 0.66 | _ |
- This table compares the performance of different predictive models in terms of Area Under the Curve (AUC) values across the training and validation datasets. The integrated model, which combines radiomics and convolutional neural networks, demonstrates superior predictive performance compared to traditional radiomics, CNN alone, and support vector machines. The p-values indicate the statistical significance of differences between the integrated model and other models, highlighting the integrated model's enhanced accuracy and generalizability in predicting the infiltration status of pulmonary ground-glass nodules (GGNs)
