Table V.
Parameter estimates and model performance statistics from multiple logistic modeling for prediction of HGPCa vs. PCa conditional on cancer being present.
| Model | Intercept (P-value) | BMI (P-value) | log(PSA) (P-value) | Age (P-value) | DRE (P-value) | Year (P-value) | Median residual | AUC ROC | Specificity (at 0.5 threshold) | Sensitivity (at 0.5 threshold) |
|---|---|---|---|---|---|---|---|---|---|---|
| Model 7 | -4.319 (<0.001) | 0.025 (0.311) | 0.562 (0.015) | 0.048 (<0.001) | 0.248 (0.229) | -0.285 (<0.001) | -0.718 | 0.675 | 0.313 | 0.856 |
| Model 8 | -3.549 (<0.001) | Omitted | 0.565 (0.015) | 0.044 (0.001) | 0.256 (0.215) | -0.285 (<0.001) | -0.728 | 0.676 | 0.294 | 0.871 |
| Model 9 | -4.058 (<0.001) | 0.026 (0.295) | Omitted | 0.049 (<0.001) | 0.371 (0.062) | -0.296 (<0.001) | -0.730 | 0.66 | 0.294 | 0.880 |
| Model 10 | -0.975 (0.176) | 0.016 (0.517) | 0.668 (0.004) | Omitted | 0.272 (0.181) | -0.295 (<0.001) | -0.760 | 0.654 | 0.229 | 0.913 |
| Model 11 | -4.412 (<0.001) | 0.026 (0.289) | 0.632 (0.005) | 0.045 (<0.001) | Omitted | -0.260 (<0.001) | -0.718 | 0.671 | 0.328 | 0.874 |
| Model 12 | -5.117 (<0.001) | 0.024 (0.330) | 0.636 (0.006) | 0.047 (<0.001) | -0.034 (0.857) | Omitted | -0.787 | 0.639 | 0.199 | 0.892 |
AUC, area under the curve; DRE, digital rectal examination; HGPCa, high-grade prostate cancer; PCa, prostate cancer; PSA, prostate specific antigen; ROC, receiver operating characteristic.