Table 2 Classification performance metrics of various machine learning algorithms.

From: Machine learning model for early prediction of acute kidney injury in heatstroke patients based on the first 24 h hospitalization data

Model

AUC (95% CI)

Accuracy (95% CI)

Precision (95% CI)

Sensitivity (95% CI)

Specificity (95% CI)

F1-score (95% CI)

Logistic regression

0.753 [0.697–0.808]

0.660 [0.603, 0.714]

0.740 [0.604, 0.860]

0.692 [0.608, 0.673]

0.922 [0.882, 0.959]

0.716 [0.634, 0.782]

SVM

0.924 [0.886–0.962]

0.879 [0.838, 0.917]

0.855 [0.793, 0.913]

0.859 [0.794, 0.921]

0.893 [0.844, 0.938]

0.857 [0.807, 0.902]

XGBoost

0.863 [0.842–0.884]

0.769 [0.744, 0.793]

0.718 [0.678, 0.758]

0.744 [0.704, 0.783]

0.788 [0.759, 0.822]

0.730 [0.697, 0.761]

KNN

0.934 [0.909–0.959]

0.841 [0.800, 0.879]

0.803 [0.733, 0.873]

0.828 [0.758, 0.891]

0.851 [0.798, 0.903]

0.814 [0.757, 0.870]

Decision tree

0.845 [0.800–0.891]

0.821 [0.779, 0.862]

0.825 [0.743, 0.898]

0.731 [0.649, 0.805]

0.886 [0.835, 0.933]

0.774 [0.714, 0.834]

Naive Bayes

0.851 [0.808–0.893]

0.624 [0.566, 0.679]

0.530 [0.465, 0.597]

0.601 [0.589, 0.621]

0.890 [0.852, 0.932]

0.691 [0.631, 0.744]

Back to article page