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

Table 3

Potential risk features ranked by weights obtained using support vector machine (SVM) feature selection [20], Blue Mountains Eye Study 10-year follow-up data.

Rank	Attribute	SVM weight
1	Age (per 1 year)	3.21660913
2	Body mass index (per 1 kg/m^2)	0.15610062
3	Current smoker (past/never smoked)	0.06839195
4	Gender (male/female)	0.05784681
5	Total cholesterol (per 1 mmol/L)	0.04203396
6	Systolic blood pressure (per 1 mmHg)	0.01872727
7	High-density lipoprotein cholesterol (per 1 mmol/L)	0.01231242
8	Diabetes (versus no diabetes)	0.00610169
9	Medication for hypertension (versus no medication for hypertension)	0.00104436
10	Retinopathy (yes/no)	0.00064500
11	Diastolic blood pressure (per 1 mmHg)	0.00023068