Predicting in-hospital mortality in patients with heart failure combined with atrial fibrillation using stacking ensemble model: an analysis of the medical information mart for intensive care IV (MIMIC-IV)

Table 2 Model performance in predicting in-hospital mortality in patients with HF combined with AF

Models	Dataset	Sensitivity (95%CI)	Specificity (95%CI)	AUC (95%CI)	Accuracy (95%CI)
Random forest	Training set	0.740 (0.706–0.775)	0.740 (0.725–0.754)	0.818 (0.801–0.835)	0.740 (0.727–0.753)
Random forest	Testing set	0.629 (0.572–0.685)	0.715 (0.692–0.738)	0.747 (0.717–0.777)	0.702 (0.681–0.723)
XGBoost	Training set	0.803 (0.772–0.834)	0.685 (0.670–0.700)	0.827 (0.811–0.843)	0.702 (0.689–0.716)
XGBoost	Testing set	0.704 (0.650–0.757)	0.670 (0.647–0.694)	0.755 (0.725–0.785)	0.676 (0.654–0.697)
LGBM	Training set	0.713 (0.678–0.749)	0.764 (0.750–0.778)	0.811 (0.794–0.829)	0.756 (0.743–0.769)
LGBM	Testing set	0.632 (0.576–0.689)	0.757 (0.735–0.778)	0.754 (0.724–0.784)	0.737 (0.717–0.758)
KNN	Training set	0.838 (0.809–0.867)	0.648 (0.633–0.664)	0.824 (0.808–0.840)	0.677 (0.662–0.691)
KNN	Testing set	0.739 (0.688–0.791)	0.637 (0.613–0.661)	0.746 (0.716–0.776)	0.653 (0.631–0.675)
Stacking ensemble	Training set	0.812 (0.782–0.843)	0.705 (0.690–0.720)	0.837 (0.821–0.852)	0.721 (0.708–0.735)
Stacking ensemble	Testing set	0.682 (0.628–0.737)	0.693 (0.670–0.716)	0.768 (0.740–0.796)	0.691 (0.670–0.712)

Note HF, heart failure; AF, atrial fibrillation; XGBoost, eXtreme Gradient Boosting; LGBM, Light Gradient Boosting Machine; KNN, K-Nearest Neighbor; AUC, the area under of the curve; CI, confidence interval

ISSN: 1472-6947