Table 5.

Predictive performance of all machine learning models

Models	Accuracy	AUC	Specificity	False negative rate	False positive rate
All features
Decision tree	0.627 (95% CI, 0.598–0.656)	0.575 (95% CI, 0.545–0.603)	0.963	0.915	0.037
Random forest	0.646 (95% CI, 0.617–0.675)	0.596 (95% CI, 0.567–0.652)	0.869	0.755	0.131
Artificial neural network	0.650 (95% CI, 0.607–0.675)	0.625 (95% CI, 0.579–0.672)	0.861	0.665	0.139
Feature selection
Decision tree	0.642 (95% CI, 0.613–0.671)	0.592 (95% CI, 0.563–0.648)	0.963	0.915	0.037
Random forest	0.648 (95% CI, 0.601–0.695)	0.605 (95% CI, 0.558 –0.652	0.913	0.802	0.087
Artificial neural network	0.668 (95% CI, 0.621–0.714)	0.654 (95% CI, 0.625–0.683)	0.922	0.755	0.078
Grace variable sets
Decision tree	0.622 (95% CI, 0.576–0.668)	0.554 (95% CI, 0.508–0.601)	0.973	0.927	0.027
Random forest	0.627 (95% CI, 0.598–0.656)	0.575 (95% CI, 0.545–0.603)	0.966	0.904	0.034
Artificial neural network	0.644 (95% CI, 0.615–0.673)	0.594 (95% CI, 0.565–0.65)	0.892	0.778	0.108