# RealICU: Do LLM Agents Understand Long-Context ICU Data? A Benchmark Beyond Behavior Imitation

arXiv: https://arxiv.org/abs/2605.13542 · HTML: https://mlsanity.com/paper/2605.13542/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Placed in Medical and scientific domain agents, one of 16 directions in LLM agents and tool use.
- Builds on 15 of its 40 references that are in the map.
- First posted on arXiv 13 May 2026.

## Placed in

- LLM agents and tool use / [Medical and scientific domain agents](https://mlsanity.com/agents/medical-and-scientific-domain-agents/)

## Builds on (most cited first, 15 of 15)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [ReAct: Synergizing Reasoning and Acting in Language Models](https://mlsanity.com/paper/2210.03629/) (2022)
- [What Disease does this Patient Have? A Large-scale Open Domain Question Answering Dataset from Medical Exams](https://mlsanity.com/paper/2009.13081/) (2020)
- [PubMedQA: A Dataset for Biomedical Research Question Answering](https://mlsanity.com/paper/1909.06146/) (2019)
- [A-MEM: Agentic Memory for LLM Agents](https://mlsanity.com/paper/2502.12110/) (2025)
- [Domain-Specific Language Model Pretraining for Biomedical Natural Language Processing](https://mlsanity.com/paper/2007.15779/) (2020)
- [The Power of Noise: Redefining Retrieval for RAG Systems](https://mlsanity.com/paper/2401.14887/) (2024)
- [AgentClinic: a multimodal agent benchmark to evaluate AI in simulated clinical environments](https://mlsanity.com/paper/2405.07960/) (2024)
- [MedXpertQA: Benchmarking Expert-Level Medical Reasoning and Understanding](https://mlsanity.com/paper/2501.18362/) (2025)
- [Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory](https://mlsanity.com/paper/2511.20857/) (2025)
- [AgentFold: Long-Horizon Web Agents with Proactive Context Management](https://mlsanity.com/paper/2510.24699/) (2025)
- [AI Hospital: Benchmarking Large Language Models in a Multi-agent Medical Interaction Simulator](https://mlsanity.com/paper/2402.09742/) (2024)
- [Yet Another ICU Benchmark: A Flexible Multi-Center Framework for Clinical ML](https://mlsanity.com/paper/2306.05109/) (2023)
- [Simulating Viva Voce Examinations to Evaluate Clinical Reasoning in Large Language Models](https://mlsanity.com/paper/2510.10278/) (2025)
- [CliBench: A Multifaceted and Multigranular Evaluation of Large Language Models for Clinical Decision Making](https://mlsanity.com/paper/2406.09923/) (2024)

## Built on by (most cited first, 3 of 3)

- [Hindsight Bias in Clinical Temporal Reasoning: How Future Data Exposure Affects Large Language Model Judgment](https://mlsanity.com/paper/2609.13454/) (2026)
- [Teaching LLMs How ICU Physicians Approach Clinical Reasoning Through OMOP-Aligned Retrieval Improves Reasoning Across Clinical Domains](https://mlsanity.com/paper/2608.22622/) (2026)
- [TRACER: A Multimodal Harness for Patient Care in Hospitals](https://mlsanity.com/paper/2610.21266/) (2026)

## Abstract (arXiv)

Intensive care units (ICU) generate long, dense and evolving streams of clinical information, where physicians must repeatedly reassess patient states under time pressure, underscoring a clear need for reliable AI decision support. Existing ICU benchmarks typically treat historical clinician actions as ground truth. However, these actions are made under incomplete information and limited temporal context of the underlying patient state, and may therefore be suboptimal, making it difficult to assess the true reasoning capabilities of AI systems. We introduce RealICU, a hindsight-annotated benchmark for evaluating large language models (LLMs) under realistic ICU conditions, where labels are created after senior physicians review the full patient trajectory. We formulate four physician-motivated tasks: assess Patient Status, Acute Problems, Recommended Actions, and Red Flag actions that risk unsafe outcomes. We partition each trajectory with 30-min windows and release two datasets: RealICU-Gold with 930-window annotations from 94 MIMIC-IV patients, and RealICU-Scale with 11,862 windows extended by Oracle, a physician-validated LLM hindsight labeler. Existing LLMs including memory-augmented ones performed poorly on RealICU, exposing two failure modes: a recall-safety tradeoff for clinical recommendations, and an anchoring bias to early interpretations of the patient. We further introduce ICU-Evo to study structured-memory agents that improves long-horizon reasoning but does not fully eliminate safety failures. Together, RealICU provides a clinically grounded testbed for measuring and improving AI sequential decision-support in high-stakes care. Project page: https://chengzhi-leo.github.io/RealICU-Bench/