Fig. 2: Comparison of the slope predictions of the generated models in the Screening Cohort.
From: Developing serum proteomics based prediction models of disease progression in ADPKD
Observed slopes were plotted against predicted slopes according to (A) Proteome, and (B) Combined Model. The solid line is the fitted using a robust linear regression methodology and the shaded area shows bootstrapped 95% confidence interval. C The difference between observed and predicted slopes according to Proteome, Combined and MIC Models were box plotted and color-coded for different CKD stages. Patient numbers for each boxplot (from left to right) are 56, 158, 55, 157, 55 and 157. RMSEs for each boxplot (from left to right) are 2.23, 1.91, 2.11, 1.83, 3.09 and 2.11. The lower and upper bounds of the box are the 25th and 75th percentile of the data, the middle line indicates the median (50th percentile) and the whiskers are 1.5 * IQR. The outliers are plotted as dots. D The plot shows the difference between observed and predicted eGFR values (ΔeGFR = Observed - Predicted) for individual patients. Data points represent individual predictions, with models color-coded as follows: Proteome Model (blue, RMSE: 11.0, n = 192), Combined Model (dark blue, RMSE: 11.0, n = 190), and Mayo Imaging Classification (MIC) Model (orange, RMSE: 12.4, n = 190). The fitted regression lines and confidence intervals illustrate the error and variability in predictions for each model. A value of ΔeGFR = 0 represents a perfect prediction, with deviations above or below indicating under- or overestimation of eGFR, respectively. The solid line is the fitted using a linear regression approach and the shaded area shows 95% confidence interval based on standard error. Source data are provided as a Source Data file.
