Table 1. Studies on predicting intravenous immunoglobulin resistance using machine learning algorithms

Ref.	N	Method	AUC	Sensitivity (%)	Specificity (%)	Accuracy (%)
21	767	Random Forest	0.916	79.7	87.3
22	109	XGBoost, etc	0.60–0.65	28–33	99–100	78–79
23	644	GBM	0.7423	30.43	99.19	88.44
24	1398	LightGBM	0.874	70.2	90.3
25	1002	LightGBM	0.78	50	88	78
26	7750	5 methods	0.626–0.711	18.1–21.6	93.3–95.0	77.6–82.1

Ref: reference; N: number of patients; AUC: area under receiver operating characteristic curve; XGBoost: extreme gradient boosting; GBM: gradient boosting machine.