arXiv：cs.LG（机器学习，全量分类）· Taye Akinrele, Noorbakhsh Amiri Golilarz, Subash Neupane, Sudip Mittal, Shahram Rahimi·· 13 小时前AI 评分33
LLM 能否进行长程推理？面向纵向临床推理的上下文策略实证评估
Can LLMs Reason Over Long Horizons? An Empirical Evaluation of Context Strategies for Longitudinal Clinical Reasoning
AI 导读
研究在 MedLoCoMo 上对比 Full、Recent、Episodic、Semantic、Hybrid 五种上下文策略与四款开源权重 LLM，考察答案正确性、对查询-证据距离的鲁棒性及无支撑前提下的拒答表现。
正文
Abstract:Longitudinal clinical reasoning requires large language models (LLMs) to identify and integrate relevant evidence distributed across extended patient histories. Although long-context models can process increasingly large amounts of information, providing more history does not necessarily make relevant evidence more accessible or improve reasoning. We compare five context strategies (Full, Recent, Episodic, Semantic, and Hybrid) on MedLoCoMo across four open-weight LLMs, examining answer correctness, robustness to query-evidence distance, and abstention on questions with unsupported premises. Episodic and Hybrid generally achieve the strongest overall accuracy, while Recent Context degrades most as supporting evidence becomes more distant; Episodic and Hybrid maintain the highest accuracy at long distances. Analysis of adversarial questions further shows that strong performance on answerable questions does not necessarily translate to successful abstention when the available history does not support the requested conclusion. These findings show that reliable longitudinal reasoning depends not only on how much history an LLM can access, but critically on how relevant evidence is selected and presented for reasoning.
| Subjects: | Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.00562 [cs.CL] |
| (or arXiv:2610.00562v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.00562 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Taye Akinrele [view email]
[v1]
Wed, 30 Sep 2026 18:36:19 UTC (362 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org