Reproduce “Learning Disease State from Noisy Ordinal Disease Progression Labels” 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

Gustav Schmidt, Holger Heidrich, Philipp Berens, Sarah Müller

The authors train a shared image encoder on retinal OCT scan pairs labeled as better, worse, stable, or other, producing an individual disease-state score for each scan. Using a small labeled set, the scores support active-versus-inactive nAMD classification. On progression classification, balanced accuracy is 59 ± 5% versus 60 ± 3% for a naïve four-class classifier, a near tie.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.