Machine learning-based prediction of postoperative pancreatic fistula after laparoscopic pancreaticoduodenectomy

BMC Surgery

Table 6 Predicted Results of Five Models for CR-POPF in Test Data

Model	Methods	Threshold	AUC	Spe	Sen	NPV	PPV
1	RF	0.725	0.604	0.853	0.385	0.879	0.333
	SVM-L	0.837	0.501	0.500	0.615	0.872	0.190
	SVM-P	0.872	0.567	0.912	0.308	0.873	0.400
	Logistic	0.170	0.563	0.691	0.538	0.887	0.250
2	RF	0.868	0.503	0.543	0.571	0.864	0.200
	SVM-L	0.840	0.502	0.800	0.357	0.862	0.263
	SVM-P	0.828	0.617	0.843	0.429	0.881	0.353
	Logistic	0.178	0.643	0.829	0.500	0.892	0.368
3	RF	0.610	0.527	0.970	0.231	0.865	0.600
	SVM-L	0.845	0.597	0.636	0.692	0.913	0.273
	SVM-P	0.846	0.607	0.545	0.769	0.923	0.250
	Logistic	0.146	0.618	0.758	0.538	0.893	0.304
4	RF	0.846	0.747^*	0.574	0.917	0.975	0.275
	SVM-L	0.853	0.609	0.471	0.917	0.970	0.234
	SVM-P	0.855	0.627	0.735	0.667	0.926	0.308
	Logistic	0.165	0.679	0.809	0.583	0.917	0.350
5	RF	0.950	0.586	0.300	0.929	0.955	0.210
	SVM-L	0.827	0.609	0.529	0.786	0.925	0.250
	SVM-P	0.866	0.639	0.400	0.929	0.966	0.236
	Logistic	0.209	0.594	0.343	0.929	0.960	0.220

Threshold, values above and equal to threshold was classed into the case group; AUC, area under the curve; Spe, specificity; Sen, Sensitivity; NPV: negative predict value; PPV: positive predict value. RF, random forest; SVM-L, support vector machine with linear kernel; SVM-P, support vector machine with polynomial kernel, Logistic, logistic regression model

ISSN: 1471-2482