Reproduce “Avoiding C-hacking when evaluating survival distribution predictions with discrimination measures” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Raphael Sonabend, Andreas Bender, Sebastian Vollmer
The authors examine comparing survival predictions when models output either risk rankings or survival curves. They recommend summing predicted cumulative hazard over time to convert a curve into expected mortality, then evaluating it with Harrell’s C. In one rats-data split, the apparent leader changes with the evaluation route; this illustration does not establish a universally best score or model.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.