Skip to main content
. 2020 Oct 26;22(10):e20891. doi: 10.2196/20891

Table 3.

Comparison results of MIMIC-III.a

Experiments AUROCb (95% CI) F1-score (95% CI) AUPRCc (95% CI) Precision (95% CI) Recall (95% CI)
SOTAd 0.857 (0.837-0.875) 0.944 (0.938-0.950) 0.505 (0.451-0.558) 0.973 (0.967-0.979) 0.773 (0.907-0.927)
Basic FLe 0.850 (0.830-0.869) 0.944 (0.938-0.950) 0.483 (0.427-0.537) 0.975 (0.969-0.980) 0.797 (0.906-0.926)
Imbalanced FL 0.850 (0.829-0.869) 0.943 (0.937-0.949) 0.481 (0.426-0.535) 0.981 (0.976-0.986) 0.714 (0.897-0.918)

aAll results are presented with a 95% CI by resampling 10,000 times.

bAUROC: area under the receiver operating characteristic curve.

cAUPRC: area under the precision-recall curve.

dSOTA: state of the art.

eFL: federated learning.