arXiv：cs.AI· Deyi Li, Qi Xu, Lingyao Li, Tiansheng Wang, Muxuan Liang, Mei Liu·· 5 小时前
ATHENA：面向电子健康记录临床预测的智能体神经架构搜索框架
Agentic neural architecture search for clinical prediction from electronic health records: Cross-hospital knowledge transfer and computational efficiency
AI 导读
ATHENA 是一个知识引导的智能体神经架构搜索框架，用于基于 Transformer 的 EHR 临床预测建模，通过每所医院预训练一次权重共享 supernet，让候选架构经微调而非独立预训练来评估，并用两层跨医院先验结合多智能体 LLM 搜索。
正文
Abstract:Transformer architectures for clinical prediction from electronic health records (EHRs) often require costly manual tuning and may vary across hospitals and tasks. We introduce ATHENA (Agentic Transfer across Hospitals for EHR Neural Architecture Search), a knowledge-guided agentic neural architecture search framework for Transformer-based EHR modeling. A weight-sharing supernet is pretrained once per hospital, allowing candidate architectures to inherit weights and be evaluated through fine-tuning rather than independent pretraining. A two-layer cross-hospital prior supplies high-performing source architecture examples and task-specific component preferences summarized by SHapley Additive exPlanations (SHAP)-based meta-regression. These priors guide a multi-agent large language model search using target-hospital validation feedback. Across six clinical prediction tasks at one held-out OneFlorida+ site and one external MIMIC-IV site, ATHENA significantly outperforms all four baselines in 9 of 12 site-task evaluations under a strict equal-compute comparison. In a separate comparison using a common pretrained AutoFormer supernet to isolate search strategy performance, ATHENA ranks first in 9 of 12 evaluations at a maximum search budget of 30 candidate evaluations. It also shows more consistent architecture selection across repeated searches. ATHENA supports more efficient architecture selection for Transformer-based clinical prediction models.
| Comments: | 37 pages, 10 figures. Revised title and presentation; clarified methods and evaluation settings; supplementary materials included |
| Subjects: | Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2608.21712 [cs.AI] |
| (or arXiv:2608.21712v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2608.21712 arXiv-issued DOI via DataCite |
Submission history
From: Deyi Li [view email]
[v1]
Sat, 22 Aug 2026 01:26:40 UTC (5,444 KB)
[v2]
Tue, 25 Aug 2026 19:53:04 UTC (5,431 KB)
[v3]
Fri, 9 Oct 2026 23:32:19 UTC (5,819 KB)
来源：arXiv：cs.AI · arxiv.org