Skip to main content
. 2020 May 21;15(5):e0232414. doi: 10.1371/journal.pone.0232414

Table 4. Comparison of the N-SRS, the R-FSRS, and other machine learning methods performance on the Validation cohort.

Reported metrics include sensitivity, specificity, precision, negative predictive value (NPV), and positive predictive value (PPV) at the probability threshold of 0.5. The overall c-statistic (AUC) and calibration χ2 results are also presented. The results refer to the aggregated population.

N-SRS R-FSRS (both genders) R-FSRS (men) R-FSRS (women) Log. Reg CART Random Forest XGBoost
Sensitivity 0.8986 0.8403 0.8411 0.8396 0.8576 0.8402 0.9055 0.9076
Specificity 0.4019 0.3663 0.3786 0.3565 0.3733 0.3599 0.4078 0.4092
Precision 0.9395 0.9320 0.9329 0.9313 0.9349 0.9348 0.9407 0.9455
NPV 0.2771 0.1815 0.1882 0.1762 0.2026 0.1805 0.2811 0.2818
PPV 0.9395 0.9320 0.9329 0.9313 0.9345 0.9317 0.9421 0.9446
AUC 0.7403 0.6491 0.6246 0.6735 0.7065 0.6829 0.7482 0.7501
AUC (95% CI) 0.7149–0.771 0.6266–0.6716 0.5931–0.6555 0.6411–0.7058 0.6772–0.7558 0.6484–0.7175 0.7198–0.7801 0.7202–0.7856
calibration χ2 7.12 36.66 37.42 35.98 25.03 35.76 6.67 6.52