Skip to main content
Fig. 4 | BMC Nephrology

Fig. 4

From: Development and validation of a machine learning model to predict time to renal replacement therapy in patients with chronic kidney disease

Fig. 4

SHAP value(a) Waterfall plot of a randomly selected case. The bottom of the waterfall plot begins with the expected value of the model output, each row shows the positive (red) or negative (blue) contribution of each item, and the top shows the final output. This patient already had severe renal dysfunction, and the expected value was 1,234.956; however, the final output was 401.096 because there were changes such as a standardized eGFR of − 695.53 and a standardized BUN of − 139.3. (b) Summary plot (scatter plot) of all cases analyzed in this study. As shown at the top of the plot, the larger and redder the eGFR feature, the larger is the positive contribution (SHAP value) to prediction, which indicates a positive correlation. (c) Summary plot (bar chart) of the average absolute SHAP values for all cases, which shows that eGFR had the greatest influence

Back to article page