Figure 2From: Diabetic retinopathy risk prediction for fundus examination using sparse learning: a cross-sectional study Performance (AUC) of the penalized logistic regression models using the 5-fold cross validation. The penalized logistic regression models included ridge (A), elastic net (B), and LASSO (C). In order to optimize λ, we investigated the AUC during the 5-fold cross validation as λ increased. The λ that indicated the highest AUC was chosen for the final training condition.Back to article page