Development of Health Parameter Model for Risk Prediction of CVD Using SVM

Table 8

Sensitivity and specificity for SVM, Framingham model, and logistic regression model with diagnostic odds ratio.

Parameter	⁢Model based on SVM classifiers		⁢Framingham risk model		⁢LRA model
	Value	95% CI	Value	95% CI	Value	95% CI
Sensitivity	0.682	0.589 to 0.764	0.52	0.4096 to 0.6275	0.48	0.3817 to 0.5809
Specificity	0.859	0.8224 to 0.89	0.70	0.6508 to 0.745	0.83	0.7926 to 0.8675
Positive likelihood ratio	4.863	3.697 to 6.396	1.73	1.326 to 2.261	2.87	2.14 to 3.85
Negative likelihood ratio	0.369	0.278 to 0.491	0.69	0.539 to 0.871	0.62	0.52 to 0.75
Diagnostic odds ratio	13.173	7.999 to 21.696	2.523	1.529 to 4.162	4.602	2.892 to 7.324
AUC	⁢0.71		⁢0.57		⁢0.63