Research Article
External Validation of an Artificial Neural Network and Two Nomograms for Prostate Cancer Detection
Table 3
ROC curve analysis for tPSA, %fPSA, ANNs, and nomograms.
| Parameter cohort | tPSA | %fPSA | Karakiewicz et al.’s Nomogram I | Kawakami et al.’s Nomogram II | ANN-Charité |
| Area under the ROC curve (AUC) |
| Saarow cohort | 0.501 | 0.669 | 0.713 | 0.742* | 0.694 | “ProstataClass” cohort | 0.7 | 0.782 | | | |
| Specificity at 95% sensitivity |
| Saarow cohort | 3.96% | 12.9% | 30.4% | 18.2% | 18.8% | “ProstataClass” cohort | 27.8% | 27.5% | | | |
| Specificity at 90% sensitivity |
| Saarow cohort | 6.93% | 25.7% | 40.9% | 27.1% | 33.7% | “ProstataClass” cohort | 39.4% | 44.1% | | | |
|
|
*Significantly different from %fPSA.
|