A non-linear ensemble model-based surgical risk calculator for mixed data from multiple surgical fields

Table 4 Results of similar models with top-15 risk factors

	Models	AUC	BS	\({\mathbf{HL}}_{{\hat{varvec{c}}}}\)	P-value
Training mode 1 (10-fold cross validation)	NL-SRC	0.892	0.063	8.082	0.426
	logit	0.864	0.068	13.561	0.094
	RF	0.881	0.066	12.281	0.139
	XGBoost	0.887	0.064	12.358	0.136
Training mode 1 (6-fold cross validation)	NL-SRC	0.890	0.064	9.753	0.283
	logit	0.867	0.068	14.603	0.067
	RF	0.882	0.066	13.372	0.100
	XGBoost	0.885	0.065	11.726	0.164
Training mode 2	NL-SRC	0.894	0.060	7.638	0.470
	logit	0.861	0.065	31.460	<0.001
	RF	0.886	0.061	14.022	0.081
	XGBoost	0.882	0.061	9.740	0.284
Training mode 3	NL-SRC	0.869	0.066	17.062	0.030
	logit	0.863	0.067	75.033	<0.001
	RF	0.864	0.069	41.323	<0.001
	XGBoost	0.874	0.069	67.251	<0.001

ISSN: 1472-6947