Maybe consider adding a parameter to use cause-specific weights (see eq 8 on [this paper](https://onlinelibrary.wiley.com/doi/abs/10.1002/bimj.201600191))?