Mortality prediction in patients with hyperglycaemic crisis using explainable machine learning: a prospective, multicentre study based on tertiary hospitals

Table 2 The values of the evaluation metrics of the models in the test set

	AUC	Accuracy	Sensitivity	Specificity	PPV	NPV
LR	0.64 (0.47, 0.79)	0.76 (0.69, 0.85)	0.20 (0.00, 0.44)	0.89 (0.81, 0.96)	0.29 (0.00, 0.60)	0.83 (0.74, 0.92)
SVM	0.70 (0.52, 0.86)	0.76 (0.65, 0.85)	0.47 (0.22, 0.73)	0.83 (0.73, 0.91)	0.39 (0.18, 0.63)	0.87 (0.77, 0.95)
RF	0.87 (0.78, 0.95)	0.80 (0.70, 0.88)	0.67 (0.42, 0.91)	0.83 (0.73, 0.91)	0.48 (0.28, 0.69)	0.92 (0.85, 0.98)
LightGBM	0.89 (0.77, 0.97)	0.83 (0.74, 0.90)	0.74 (0.47, 0.94)	0.85 (0.76, 0.93)	0.52 (0.31, 0.74)	0.94 (0.87, 0. 99)
DNN	0.64 (0.54, 0.87)	0.81 (0.73, 0.89)	0.26 (0.06, 0.53)	0.94 (0.88, 0.99)	0.50 (0.11, 0.88)	0.85 (0.76, 0.92)

LR logical regression, SVM support vector machine, RF random forest, LightGBM light gradient boosting machine, DNN deep neural network algorithm, NPV negative predictive value, PPV positive predictive value

ISSN: 1758-5996