Development of a Bagged CART Model for Subclassification of Diabetic Retinopathy Using Metabolomics Data

Abstract

In this study, we present the development and evaluation of a predictive model for classifying the subclasses of diabetic retinopathy—No Diabetic Retinopathy (NDR), Non-Proliferative Diabetic Retinopathy (NPDR), and Proliferative Diabetic Retinopathy (PDR)—using metabolomics data. The metabolomics dataset underwent rigorous preprocessing to address missing values, employing the Random Forest algorithm, and was subsequently normalized to ensure comparability across all samples. A bagged Classification and Regression Trees (CART) algorithm was utilized to construct the prediction model, leveraging its robustness and accuracy for classification tasks. Our model demonstrated significant potential in accurately classifying diabetic retinopathy subclasses, suggesting that metabolomics data, when combined with advanced machine learning techniques, can provide valuable insights into the progression and management of diabetic retinopathy. This study underscores the importance of integrating metabolomics biomarkers and machine learning for the advancement of personalized medicine in diabetic care.