problem for the network's structure

[Ge-YL]

Hi, Could you answer me a questions? I come meet the problem that all patients' risk is a same value with the update of the network in the training time. Is the network too simple can explain the phenomenon?

[carped1um]

1) Due to the complex pipeline for multimodal cancer risk prediction, I cannot reply your questions with confidence. 2) As for the issue about if the framework is too simple, you can justify it by constructing a substantial simple network, e.g., MLP, for preliminary experiment. This is a feasible way to answer your question.^_^ 3) Frankly, such triple-level fusion framework may be too complicated…… Good Luck

[Ge-YL]

1. Due to the complex pipeline for multimodal cancer risk prediction, I cannot reply your questions with confidence.由于多模式癌症风险预测的管道复杂，我无法自信地回答您的问题。
2. As for the issue about if the framework is too simple, you can justify it by constructing a substantial simple network, e.g., MLP, for preliminary experiment. This is a feasible way to answer your question.^^至于框架是否太简单的问题，你可以通过构建一个实质性的简单网络来证明它的合理性，例如 MLP，进行初步实验。这是回答您的问题的可行方法。 ^^
3. Frankly, such triple-level fusion framework may be too complicated……坦率地说，这样的三级融合框架可能太复杂了...... Good Luck 祝你好运

Perhaps our data preprocessing is different. In addition, I only used your bimodal network.

[Ge-YL]

1. Due to the complex pipeline for multimodal cancer risk prediction, I cannot reply your questions with confidence.由于多模式癌症风险预测的管道复杂，我无法自信地回答您的问题。
2. As for the issue about if the framework is too simple, you can justify it by constructing a substantial simple network, e.g., MLP, for preliminary experiment. This is a feasible way to answer your question.^^至于框架是否太简单的问题，你可以通过构建一个实质性的简单网络来证明它的合理性，例如 MLP，进行初步实验。这是回答您的问题的可行方法。 ^^
3. Frankly, such triple-level fusion framework may be too complicated……坦率地说，这样的三级融合框架可能太复杂了...... Good Luck 祝你好运

Can you share the download link for BFTools used in your article? I have currently found several BFtools on GitHub, but I cannot determine which one it is.

[carped1um]

I would like to support this, but the data preprocessing is conducted by a co-author... I think that u may find the result accoding to the refernce "[38] Sun et al., Integrating genomic data and pathological images to effectively predict breast cancer survival outcome, CMPB, 2018". It is a fundamental work of my lab earlier. Best wishes