Table 2 Predictive performance comparison in the test set for aggregated and individual models, BP Hospital—A Beneficência Portuguesa de São Paulo, Brazil, 2020.

ICU + MV

Predict ICU

Random forest

0.959 [0.94; 098]

0.906

0.868

0.720

0.961

0.802

Predict MV

0.912 [0.87; 0.95]

0.935

0.723

0.271

0.990

0.420

Predict death

0.925 [0.89; 0.96]

0.969

0.730

0.290

0.995

0.446

Only death

Predict death

Extra trees

0.972 [0.95; 1.00]

0.964

0.863

0.409

0.996

0.574

ICU + death

Predict ICU

XGBoost

0.965 [0.95; 0.98]

0.847

0.930

0.818

0.942

0.832

Predict MV

0.925 [0.89;0.96]

0.946

0.808

0.398

0.991

0.560

Predict Death

0.922 [0.89; 0.95]

1.000

0.787

0.307

1.000

0.470

Only MV

Predict MV

Extra trees

0.945 [0.91;0.98]

0.906

0.819

0.362

0.987

0.518

MV + death

Predict ICU

Random forest

0.921 [0.89; 0.95]

0.765

0.901

0.729

0.917

0.747

Predict MV

0.940 [0.91; 0.97]

0.933

0.799

0.329

0.991

0.487

Predict death

0.943 [0.91; 0.98]

0.963

0.794

0.306

0.996

0.464

Only ICU

Predict ICU

Random forest

0.959 [0.94; 0.98]

0.906

0.868

0.720

0.961

0.802