From: Utilizing imbalanced electronic health records to predict acute kidney injury by ensemble learning and time series model
AUC
Sensitivity
F1-socre
AP
24h
48h
Random Undersample
0.81
0.78
0.75
0.68
0.58
0.44
0.59
0.41
Random Oversample
0.69
0.64
0.43
0.62
0.66
0.46
Cost-sensitive XGBoost
0.70
0.45
0.61
0.67
0.47