EarlyDx: An Admission-Anchored Benchmark for Open-Ended Generation of Evidence-Supported ED-Encounter Diagnoses
New open-ended benchmark exposes diagnostic limitations of LLMs during critical emergency department admissions.
Clinical diagnosis at hospital admission must be made rapidly from limited, incomplete evidence. Existing diagnosis-prediction benchmarks are poorly suited to this setting: they restrict prediction to closed code sets, exclude free-text notes, and supervise with discharge diagnos...
ArXiv CS.AI
Source
Generated insight by OnAirToday AI
EarlyDx addresses a critical gap in clinical AI by shifting the focus from post-discharge outcomes to the high-stakes, data-sparse environment of emergency department admissions. By forcing models to rely on limited initial evidence, the benchmark highlights a significant performance ceiling in current reasoning capabilities for urgent diagnostic tasks.

Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.