Metraipolis › archive › dc_529e13d69571
Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents — arXiv:2609.13543
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.13543
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T22:46:48.029Z
- SHA-256 of the archived text
826bf66e886e950019cc59e4ceebfb0be1d4109c9d27e3969af0c4f5920f7391- Stable URL of this page
https://metraipolis.com/documento/dc_529e13d69571
How to cite
Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents — arXiv:2609.13543 — arXiv (en), https://arxiv.org/abs/2609.13543 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_529e13d69571 (Metraipolis archive dc_529e13d69571, sha256 826bf66e886e950019cc59e4ceebfb0be1d4109c9d27e3969af0c4f5920f7391)
BibTeX
@misc{dc529e13d69571,
title = {Asclepius: An Adaptive Harness for Long-Horizon Clinical Agents — arXiv:2609.13543},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.13543},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_529e13d69571; sha256 826bf66e886e950019cc59e4ceebfb0be1d4109c9d27e3969af0c4f5920f7391},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.13543. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM agents are predominantly benchmarked on short, single-task trajectories, yet real deployments run for hours under contention, surfacing a different class of failures. We use the Clinical Environment Simulator (CES), in which an agent manages an entire emergency-department shift under continuous time and resource pressure, as a testbed: long-horizon execution failures manifest measurably in a single rollout under structured, multi-dimensional grading. On CES, current agents reach the correct diagnosis in most cases yet fail to deliver complete and timely critical actions, revealing an execution gap. We attribute this gap to three long-horizon failure modes, each operationalized as a per-trace counter: instruction-adherence drift, treatment incompleteness, and a severity-equity gap in timeliness. We then introduce Asclepius, an adaptive agent scaffolding with a self-evolving harness that rewrites the operating manual between shifts from trace-level feedback, an externalized clinical skills library for high-stakes regimen knowledge, and three isolated subagents that partition per-turn decisions across the patient queue. On held-out batches never observed during harness evolution, Asclepius improves critical-action correctness by 22% (p = 0.024) over a strong baseline agent framework while preserving diagnostic accuracy, with consistent gains across five LLM judges from three model families; on the full ten-batch set, improvements reach 25% on critical actions and 13% on timeliness. The three failure modes form a coupled bottleneck: decisive reductions appear only when all three components act together. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-15 in cs.AI). Categorie arXiv: Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2609.13543 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-15T22:29:03.076Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-15?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).