Reply to: Clarifying multimodal inputs and attribution in colorectal cancer survival prediction

We thank Wang et al. for their interest in our work and for raising these insightful points. In this response, we first clarify the specific input modalities utilized for survival prediction during the inference phase. Furthermore, we provide a precise definition of the comparative baselines referenced, detailing how multi-source data contributes to the training process. Finally, we elucidate the internal data flow within our proposed framework, specifically addressing which feature representations are fed into the survival analysis head.