Figure 2.

Performance of the model for AMD lifetime risk assessment based on 27 variables (genetic and environmental factors). (a) Receiver operating characteristic curve (ROC) for the 25-SNP plus smoking and BMI (body mass index) model was generated for learning (blue line) and testing (red dashed line) sets by using the binary logistic regression analysis with a 10-fold cross-validation method. (b) Prediction success parameters were calculated for testing set. (c) Variable importance ranking is showing relative scores generated by TreeNet software and positioning the predictors from the most important to the least important. AMD, age-related macular degeneration cases; NC, normal control; NPV, negative prediction value; PPV, positive prediction value.