Skip to main content
. 2025 Oct 13;12:1655302. doi: 10.3389/fmed.2025.1655302

FIGURE 4.

Scatter plot showing SHAP feature importance for gold prediction. Features on the y-axis include scarheight, age, S_mean, and Solidity. The x-axis represents SHAP values, showing the impact on model output, ranging from negative to positive. Data points are colored by feature value, from low (blue) to high (pink).

SHAP Summary plot of feature importance for prediction of scar outcome using random forest. The SHAP beeswarm plot ranks the top four predictors according to their overall impact on model output. Each point represents an individual case, with horizontal position denoting the SHAP value (the impact of the variable on the predicted outcome) and color indicating the feature value [from low (blue) to high (red)]. Solidity and S_mean show negative SHAP values for higher feature values, indicating protective effects against poor outcome. In contrast, higher scar height and age are associated with increased SHAP values, reflecting greater risk of poor prognosis. These findings confirm the importance and independent contributions of both image-based and clinical features in predicting scar treatment response.