Table 2.

Performance of the six ML models for the estimation of mortality of patients with a POCA.

Models	AUC [95%CI]	Accuracy	Precision	Recall	f1-score
Logistic regression	0.84 [0.71–0.95]	0.74	0.78	0.70	0.74
Support vector classifier	0.87 [0.73–0.96]	0.79	0.83	0.75	0.79
Random forest	0.91 [0.79–0.98]	0.82	0.84	0.80	0.82
Gradient boost machine	0.90 [0.79–0.98]	0.82	0.84	0.80	0.82
Adaptive boosting classifier	0.87 [0.73–0.97]	0.76	0.79	0.75	0.77
Ensemble (VotingClassifer)	0.90 [0.78–0.98]	0.84	0.85	0.85	0.85

The 95% CI of AUC was calculated from 1000 bootstrap resamples of predictions on the test data.