Research Papers 论文研究 7h ago Updated 2h ago 更新于 2小时前 45

Standalone LLM and a Pre-specified Agentic Pipeline for Explaining ICU Mortality Predictions: a Feasibility Study on the eICU Demo Dataset 独立大语言模型与预定义智能体管道用于解释ICU死亡率预测:基于eICU演示数据集的可行性研究

Standalone LLM outperforms agentic pipeline in SHAP alignment (Jaccard 0.171 vs 0.077) and direction consistency (92.9% vs 78.6%) when explaining ICU mortality predictions Agentic pipeline demonstrates superior guideline grounding (0.762 vs 0.143), value specificity (0.236 vs 0.143), and slightly higher plausibility (0.700 vs 0.671) Standalone LLM produced 1 explanation with explicit outcome leakage on a 38-case subset, while the four-step agentic pipeline produced none XGBoost baseline achieved 研究比较了独立LLM与四步Agentic Pipeline在解释ICU死亡率预测中的效果,使用eICU Demo数据集(2,353例,死亡率8.1%) XGBoost预测模型达到AUROC 0.855、AUPRC 0.332,为后续解释提供基础 独立LLM在SHAP对齐度(Jaccard 0.171 vs 0.077)和方向一致性(92.9% vs 78.6%)上更优,但出现1例结果泄露 Agentic Pipeline在指南 grounding(0.762 vs 0.143)、值特异性(0.236 vs 0.143)和合理性(0.700 vs 0.671)上更优,且无结果泄露 临床建议:Ag

58
Hot 热度
72
Quality 质量
65
Impact 影响力

Analysis 深度分析

TL;DR

  • Standalone LLM outperforms agentic pipeline in SHAP alignment (Jaccard 0.171 vs 0.077) and direction consistency (92.9% vs 78.6%) when explaining ICU mortality predictions
  • Agentic pipeline demonstrates superior guideline grounding (0.762 vs 0.143), value specificity (0.236 vs 0.143), and slightly higher plausibility (0.700 vs 0.671)
  • Standalone LLM produced 1 explanation with explicit outcome leakage on a 38-case subset, while the four-step agentic pipeline produced none
  • XGBoost baseline achieved AUROC of 0.855 and AUPRC of 0.332 on the eICU Demo dataset (2,353 ICU stays, 8.1% mortality)
  • Agentic decomposition improves safety-relevant grounding but should be paired with attribution-based checks for high-stakes clinical risk explanation

Why It Matters

This study directly addresses a critical gap in clinical AI deployment: the disconnect between accurate predictive models and the narrative explanations clinicians need at the bedside. By comparing standalone LLMs against multi-step agentic pipelines, it provides actionable evidence for healthcare AI practitioners deciding between simplicity and structured reasoning in high-stakes medical explanation systems.

Technical Details

  • Dataset: Retained local eICU Demo artifact set comprising 2,353 ICU stays with an 8.1% mortality rate; a stratified 38-case explanation subset was used for LLM evaluation, with 14 cases overlapping a SHAP review subset
  • Baseline model: XGBoost classifier achieving AUROC 0.855 (95% CI 0.796–0.906) and AUPRC 0.332 (95% CI 0.217–0.494) for ICU mortality prediction
  • Standalone LLM: Direct explanation generation without intermediate reasoning steps; evaluated on SHAP alignment, direction consistency, and outcome leakage
  • Four-step agentic pipeline: Decomposed pipeline separating data interpretation, guideline checking, and final explanation generation; evaluated on guideline grounding, value specificity, and plausibility metrics
  • Evaluation metrics: Jaccard similarity for feature alignment with SHAP, direction consistency percentage, guideline grounding score, value specificity, plausibility rating, and outcome leakage detection

Industry Insight

  • Healthcare AI systems should consider hybrid approaches that combine the attribution fidelity of standalone LLMs with the safety grounding of agentic pipelines, rather than treating them as mutually exclusive architectures
  • Outcome leakage remains a critical validation concern for standalone LLMs in clinical settings; agentic decomposition with explicit guideline-checking steps can serve as an effective safeguard
  • The modest plausibility difference (0.700 vs 0.671) suggests that pipeline complexity does not automatically translate to meaningfully better clinical explanations, warranting cost-benefit analysis before deploying multi-step agents in production healthcare environments

TL;DR

  • 研究比较了独立LLM与四步Agentic Pipeline在解释ICU死亡率预测中的效果,使用eICU Demo数据集(2,353例,死亡率8.1%)
  • XGBoost预测模型达到AUROC 0.855、AUPRC 0.332,为后续解释提供基础
  • 独立LLM在SHAP对齐度(Jaccard 0.171 vs 0.077)和方向一致性(92.9% vs 78.6%)上更优,但出现1例结果泄露
  • Agentic Pipeline在指南 grounding(0.762 vs 0.143)、值特异性(0.236 vs 0.143)和合理性(0.700 vs 0.671)上更优,且无结果泄露
  • 临床建议:Agentic分解可提升安全性相关依据和患者特异性细节,但高风险场景需配合归因检查使用

为什么值得看

本文为医疗AI可解释性提供了独立的LLM与Agentic Pipeline对比实证,揭示了"准确性对齐"与"临床安全性"之间的权衡。对从事医疗AI落地、可解释性研究或临床决策支持系统的从业者具有重要参考价值。

技术解析

  • 数据集与模型:使用eICU Demo数据集(2,353 ICU住院记录,死亡率8.1%),XGBoost作为基线预测模型,AUROC 0.855(95% CI 0.796-0.906),AUPRC 0.332(95% CI 0.217-0.494)
  • 独立LLM方案:直接使用大语言模型生成ICU死亡率预测解释,在38例分层子集上测试,与SHAP审查子集(14例)对比评估
  • 四步Agentic Pipeline:将解释流程分解为数据解释、指南检查、最终解释等步骤,分离处理以降低风险
  • 评估指标:SHAP对齐度(Jaccard指数)、方向一致性、指南grounding、值特异性、合理性评分,以及结果泄露检测

行业启示

  • 可解释性架构设计:Agentic Pipeline通过流程分解提升了临床安全性和指南遵循度,但牺牲了与特征归因的一致性,提示医疗AI系统需在"解释准确性"与"临床安全性"间权衡
  • 高风险场景的混合策略:研究建议将Agentic分解与归因检查结合使用,为医疗AI落地提供了"多方法融合"的实践路径
  • 结果泄露风险管控:独立LLM出现结果泄露问题,凸显了医疗AI解释系统中数据隔离和验证机制的重要性,行业需建立严格的解释质量审查流程

Disclaimer: The above content is generated by AI and is for reference only. 免责声明:以上内容由 AI 生成,仅供参考。

LLM 大模型 Agent Agent Healthcare AI 医疗AI Research 科学研究 Dataset 数据集