Table 3 Model performance for each algorithm and outcome

Random forests

 AUROC

0.942 [0.925, 0.958]

0.945 [0.928, 0.960]

0.862 [0.819, 0.902]

0.851 [0.807, 0.893]

 Calibration

1.414 [1.203, 1.701]

1.374 [1.161, 1.642]

0.935 [0.778, 1.131]

0.894 [0.743, 1.090]

 Low sensitivity

0.988

0.967

0.988

0.953

 High PPV

0.865

0.856

0.355

0.461

LASSO

 AUROC

0.911 [0.890, 0.930]

0.930 [0.911, 0.949]

0.821 [0.765, 0.872]

0.860 [0.820, 0.898]

 Calibration

0.926 [0.807, 1.082]

1.071 [0.946, 1.245]

0.673 [0.527, 0.848]

0.826 [0.680, 0.999]

 Low sensitivity

0.979

0.935

0.976

0.941

 High PPV

0.838

0.786

0.335

0.434

LASSO interactions

 AUROC

0.917 [0.897, 0.936]

0.932 [0.911, 0.950]

0.838 [0.795, 0.879]

0.860 [0.817, 0.898]

 Calibration

0.980 [0.853, 1.146]

0.993 [0.871, 1.166]

0.813 [0.671, 0.9876]

0.952 [0.789, 1.138]

 Low sensitivity

0.976

0.885

0.976

0.918

 High PPV

0.845

0.769

0.346

0.418