arXiv cs.AI· Alexander Romanus, Mehul Shetty, Connor Jordan·· 13 小时前AI 评分24
利用 CausaLM 量化临床决策中的症状因果性
Quantifying Symptom Causality in Clinical Decision Making: An Exploration Using CausaLM
AI 导读
研究基于 CausaLM 框架生成反事实文本表示，使目标概念被“遗忘”，从而估计“胸痛”等症状对模型诊断预测的因果影响。团队用 TReATE 量化症状存在与否如何改变模型输出的疾病分布，并与 CONEXP 等基于相关性的基线对比。结果揭示临床 NLP 模型的决策行为，为更可信、可解释的因果决策支持工具提供依据。
正文 · 原文
Abstract:Current machine learning approaches to medical diagnosis often rely on correlational patterns between symptoms and diseases, risking misdiagnoses when symptoms are ambiguous or common across multiple conditions. In this work, we move beyond correlation to investigate the causal influence of key symptoms-specifically "chest pain" on diagnostic predictions. Leveraging the CausaLM framework, we generate counterfactual text representations in which target concepts are effectively "forgotten" enabling a principled estimation of the causal effect of that concept on a model's predicted disease distribution. By employing Textual Representation-based Average Treatment Effect (TReATE), we quantify how the presence or absence of a symptom shapes the model's diagnostic outcomes, and contrast these findings against correlation-based baselines such as CONEXP. Our results offer deeper insight into the decision-making behavior of clinical NLP models and have the potential to inform more trustworthy, interpretable, and causally-grounded decision support tools in medical practice.
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2503.19394 [cs.LG] |
| (or arXiv:2503.19394v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2503.19394 arXiv-issued DOI via DataCite |
Submission history
From: Mehul Shetty [view email]
[v1]
Tue, 25 Mar 2025 06:59:21 UTC (238 KB)
[v2]
Wed, 21 Oct 2026 03:40:42 UTC (236 KB)
来源：arXiv cs.AI · arxiv.org