Random forest can accurately predict the development of end-stage renal disease in immunoglobulin a nephropathy patients

Abstract

Background IgA nephropathy (IgAN) is the most common glomerulonephritis worldwide and up to 40% will develop end-stage renal disease (ESRD) within 20 years. However, predicting which patients will progress to ESRD is difficult. The purpose of this study was to develop a predictive model which could accurately predict whether IgAN patients would progress to ESRD. Methods Six machine learning algorithms were used to predict whether IgAN patients would progress to ESRD: logistic regression, random forest, support vector machine (SVM), decision tree, artificial neural network (ANN), k nearest neighbors (KNN). Nineteen demographic, clinical, pathologic and treatment parameters were used as input for the prediction models. Results Random forest is best able to predict progression to ESRD. The model had accuracy of 93.97% and sensitivity and specificity of 80.60% and 95.27%, respectively. Conclusions Machine learning algorithms can effectively predict which patients with IgA nephropathy will progress to end stage renal disease.