I$\text{'}$m working with the torchsurv library to implement a Convolutional Neural Network $($CNN$)$ for survival analysis, and I need assistance with calculating the C$-$index. I have already split my data into two scenarios: train$-$test and train$-$validation$-$test. My main challenges are:

C$-$Index Calculation:

How do I calculate the C$-$index for a CNN model in torchsurv when using a train$-$test split?

What adjustments are needed in the code to calculate the C$-$index when using a train$-$validation$-$test split?

Model Evaluation:

How should I structure my CNN model in torchsurv to ensure accurate C$-$index calculation?

What is the correct way to use torch.no$\_$grad during evaluation to compute the C$-$index without affecting the model's gradients?

Any detailed guidance or example code snippets for these scenarios would be greatly appreciated. Thank you!