Table 1 Performance of the machine learning models based on the test set.

From: Development and external validation of an interpretable machine learning model for the prediction of intubation in the intensive care unit

Model

AUC (95% CI)

Accuracy (95% CI)

Sensitivity (95% CI)

Specificity (95% CI)

F1_score

MCC

CatBoost

0.881

(0.856–0.905)

0.762

(0.745–0.780)

0.843

(0.780–0.902)

0.758

(0.706–0.742)

0.284

(0.241–0.325)

0.309

(0.268–0.348)

LightGBM

0.873

(0.849–0.897)

0.760

(0.743–0.776)

0.828

(0.763–0.890)

0.756

(0.704–0.741)

0.279

(0.237–0.316)

0.300

(0.257–0.336)

XGBoost

0.871

(0.843–0.895)

0.755

(0.737–0.773)

0.806

(0.736–0.872)

0.752

(0.702–0.769)

0.269

(0.230–0.309)

0.286

(0.245–0.328)

Random forest

0.824

(0.791–0.854)

0.736

(0.717–0.754)

0.806

(0.740–0.874)

0.731

(0.682–0.721)

0.254

(0.215–0.295)

0.270

(0.228–0.311)

SVM

0.813

(0.778–0.845)

0.578

(0.570–0.610)

0.888

(0.833–0.937)

0.560

(0.527–0.568)

0.191

(0.165–0.223)

0.206

(0.179–0.240)

Logistic regression

0.813

(0.779–0.848)

0.576

(0.558–0.590)

0.881

(0.818–0.935)

0.558

(0.514–0.573)

0.188

(0.158–0.217)

0.202

(0.169–0.231)

Naïve Bayes

0.768

(0.729–0.805)

0.548

(0.528–0.577)

0.813

(0.746–0.867)

0.532

(0.492–0.563)

0.167

(0.143–0.193)

0.159

(0.126–0.192)

Back to article page