. Author manuscript; available in PMC: 2012 Oct 1.

Published in final edited form as: J Biomed Inform. 2011 May 27;44(5):859–868. doi: 10.1016/j.jbi.2011.05.004

Table 5.

Comparative performance of the weighted BNI, the conventional BNI, KNN and SVM: the weighted BNI model is 10% more accurate than the conventional BNI model for pancreatic cancer prediction.

	Sensitivity	Specificity	Accuracy	ROC
	Sensitivity	Specificity	Accuracy	AUC	SE	95% CI
Weighted BNI	0.847	0.852	0.850	0.910	0.021	(0.869, 0.951)
Conventional BNI	0.796	0.704	0.735	0.806	0.029	(0.750, 0.863)
KNN	0.650	0.332	0.389	0.718	0.046	(0.654, 0.776)
SVM	0.768	0.396	0.534	0.727	0.043	(0.663, 0.785)