Table 2.
Summary of the performances of the breast cancer recurrence prediction models on the fine-tuning dataset in terms of accuracy, sensitivity, and specificity, using the Dynamically Selected Features (DSF), the Optimal Subset of Features (OSF), also combined with the clinical features (OSF + clinical), respectively. The features were selected for the exams at timepoint T1 alone or at timepoints T1–T2 simultaneously (original model). The numbers of features selected for OSF and OSF + clinical are also highlighted. The numbers of DSF features are omitted because they vary from one patient to another.
| Fine-Tuning Dataset | ||||||
|---|---|---|---|---|---|---|
| Performance Metric | DSF | OSF | OSF + Clinical | |||
| T1 | T1–T2 | T1 | T1–T2 | T1 | T1–T2 | |
| N. features | - | - | 15 | 13 + 5 | 15 + 4 | 13 + 5 + 4 |
| Accuracy | 67.7% | 72.9% | 82.3% | 87.5% | 87.5% | 91.7% |
| Sensitivity | 38.5% | 57.7% | 57.7% | 80.8% | 69.2% | 80.8% |
| Specificity | 78.6% | 78.6% | 91.4% | 90.0% | 94.3% | 95.7% |