fig7

Figure 7. SHAP analysis of NoGPF and GPFOnly-trained model on both source and target domains. (A-C) Mean absolute SHAP value of each feature in NoGPF-trained model on the (A) source and (B) target, with (C) the deviation between two domains; (D-F) Mean absolute SHAP value of each feature in GPFOnly-trained model on the (D) source and (E) target, with (F) the deviation between two domains. SHAP: Shapley additive explanations.