Reproduce “Rethinking Correctness for Uncertainty Estimation in Clinical Prediction with Vision-Language Models” 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
Mingcheng Zhu, Jinning Liang, Tingting Zhu
The authors test how correctness rules affect evaluation of uncertainty methods for clinical model predictions. They compare eight rules with human judgments and with uncertainty results referenced to those judgments, using a 450-prediction audit and a 23,254-prediction set. Exact matching reached 0.992 balanced accuracy and 0.005 AUROC distortion, but full-set AUROC shifted 0.146 and rankings reversed; human labels were unavailable there.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.