Table 4 Biopsies avoided and test trade-off results for the constructed models to predict csPCa using different risk thresholds
Different risk thresholds | Clinical model | bpMRI PI-RADS v2.1-based model | ||||
|---|---|---|---|---|---|---|
Overall biopsies avoided | Detected csPCa without unnecessary biopsies | Test trade-off, patients biopsied per detected csPCa | Overall biopsies avoided | Additional csPCa detected (without a change in unnecessary biopsies) when using the bpMRI PI-RADS v2.1-based model rather than the clinical model | Test trade-off, patients undergoing bpMRI per avoided unnecessary biopsy | |
Development Cohort | ||||||
 10% | 0 | 23.2 | 4.3 | 26.1 | 2.6 | 3.8 |
 15% | 13.6 | 20.1 | 4.9 | 31.8 | 3.5 | 2.8 |
 20% | 11.2 | 15.1 | 6.6 | 36.0 | 6.5 | 1.5 |
Validation Cohort 1 | ||||||
 10% | 0 | 25.9 | 3.8 | 15.9 | 1.8 | 5.5 |
 15% | 0 | 21.6 | 4.6 | 24.5 | 3.6 | 2.7 |
 20% | 0 | 10.9 | 9.1 | 31.3 | 14 | 0.7 |
Validation Cohort 2 | ||||||
 10% | 0 | 34.9 | 2.8 | 0 | 0 | 0 |
 15% | 0 | 31.1 | 3.2 | 5.3 | 0.9 | 9.1 |
 20% | 0 | 26.8 | 3.7 | 11.6 | 2.8 | 3.5 |