Figure 3.

Validation of the machine learning models. (A) Pie charts depict the accuracy and precision achieved in differentiating kidney diseases with the use of machine learning models. The left panels present the results of models that used only standard clinical laboratory data. The right panels present the results of models that added the urinary sphingolipid measurements. The top 10–20 important variables for prediction are shown for (B) the models using only standard clinical laboratory data and (C) the models that added the urinary sphingolipid measurements.