Figure 1.

Summary of the SHAP values based on the random forest classifier model. The y-axis shows the most relevant features for predicting graft detachment, from highest to lowest. The x-axis displays the SHAP values, reflecting the effect of each variable on the model’s outcome by indicating the predicted change in the probability of a graft detachment. A negative SHAP value represents a “protective effect” (i.e., decreased risk of detachment), while a positive SHAP value represents a “risk effect” (i.e., increased risk of detachment). Each symbol in the plot represents an individual patient in the test dataset. The color gradient ranging from blue to red represents the range of values of the variables. For binary variables, only two colors are used (red for “yes” and blue for "no"); for continuous variables, the values are depicted using the entire spectrum from red to blue.