Reproduce “Rethinking Next-Visit Diagnosis Prediction Beyond Aggregate Metrics: Retrieval--Memory Multi-Agent Adjudication” 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
Heejoon Koo
The authors introduce LONGINUS, a training-free system for predicting diagnosis codes at a patient’s next visit. It combines retrieval from visit-aware history with a trajectory memory, then adjudicates disagreements without revealing which view proposed the diagnosis. Across two MIMIC datasets and four LLM backbones, it achieves best transition-specific F1 in 22 of 24 settings, but does not lead uniformly on aggregate metrics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.