Hi, I just got ideas from your paper and code: i used my own model which accepts two modality image mri and pet to perform model_wrappers in your code. And I got nice AUC、PR curves, but when i started to do shap analysis, the shap_values array constains a lots of zero. This is my first time using shap analysis, so I want to know what causes such problems.
looking forward to your reply!
Yueup
commented
1 year ago
Hi, I just got ideas from your paper and code: i used my own model which accepts two modality image mri and pet to perform model_wrappers in your code. And I got nice AUC、PR curves, but when i started to do shap analysis, the shap_values array constains a lots of zero. This is my first time using shap analysis, so I want to know what causes such problems. looking forward to your reply!
您好，受了您的论文和代码启发之后，我尝试自己构建的模型使用MRI与PET两种模态影像联合诊断疾病阶段，在ROC与PR结果方面表现很好，但是我在进行shap分析生成于FigureTable/NeuroPathRegions/shap_csvfiles中的csv文件包含了很多0值，这导致我在后续进行您代码中FigureTable/brainNetwork时出现了问题*，所以我想知道在模型表现优异的情况下，shap分析得到的结果不理想的原因是什么。 期待您的回复!