Fig. 5From: Machine-learning-based models to predict cardiovascular risk using oculomics and clinic variables in KNHANESROC and AUC of models use both oculomics and clinics as input variables. a-b The performance of our model using AIP=0.318 or 0.34 as the cut-off point in the internal validation dataset is demonstrated. c-d The performance of our model using AIP=0.318 or 0.34 as the cut-off point in the external validation dataset is demonstrated. AIP, Atherogenic index of plasma. AUC, Area under curve. ROC, Receiver operating characteristicBack to article page