@inproceedings{sakib-etal-2025-spurious,
title = "Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in {SDOH} Extraction with Large Language Models",
author = "Sakib, Fardin Ahsan and
Zhu, Ziwei and
Grace, Karen Trister and
Yetisgen, Meliha and
Uzuner, Ozlem",
editor = "Che, Wanxiang and
Nabende, Joyce and
Shutova, Ekaterina and
Pilehvar, Mohammad Taher",
booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.acl-short.86/",
doi = "10.18653/v1/2025.acl-short.86",
pages = "1097--1106",
ISBN = "979-8-89176-252-7",
abstract = "Social determinants of health (SDOH) extraction from clinical text is critical for downstream healthcare analytics. Although large language models (LLMs) have shown promise, they may rely on superficial cues leading to spurious predictions. Using the MIMIC portion of the SHAC (Social History Annotation Corpus) dataset and focusing on drug status extraction as a case study, we demonstrate that mentions of alcohol or smoking can falsely induce models to predict current/past drug use where none is present, while also uncovering concerning gender disparities in model performance. We further evaluate mitigation strategies{---}such as prompt engineering and chain-of-thought reasoning{---}to reduce these false positives, providing insights into enhancing LLM reliability in health domains."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sakib-etal-2025-spurious">
<titleInfo>
<title>Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in SDOH Extraction with Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Fardin</namePart>
<namePart type="given">Ahsan</namePart>
<namePart type="family">Sakib</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ziwei</namePart>
<namePart type="family">Zhu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Karen</namePart>
<namePart type="given">Trister</namePart>
<namePart type="family">Grace</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Meliha</namePart>
<namePart type="family">Yetisgen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ozlem</namePart>
<namePart type="family">Uzuner</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Wanxiang</namePart>
<namePart type="family">Che</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Joyce</namePart>
<namePart type="family">Nabende</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ekaterina</namePart>
<namePart type="family">Shutova</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mohammad</namePart>
<namePart type="given">Taher</namePart>
<namePart type="family">Pilehvar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Vienna, Austria</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-252-7</identifier>
</relatedItem>
<abstract>Social determinants of health (SDOH) extraction from clinical text is critical for downstream healthcare analytics. Although large language models (LLMs) have shown promise, they may rely on superficial cues leading to spurious predictions. Using the MIMIC portion of the SHAC (Social History Annotation Corpus) dataset and focusing on drug status extraction as a case study, we demonstrate that mentions of alcohol or smoking can falsely induce models to predict current/past drug use where none is present, while also uncovering concerning gender disparities in model performance. We further evaluate mitigation strategies—such as prompt engineering and chain-of-thought reasoning—to reduce these false positives, providing insights into enhancing LLM reliability in health domains.</abstract>
<identifier type="citekey">sakib-etal-2025-spurious</identifier>
<identifier type="doi">10.18653/v1/2025.acl-short.86</identifier>
<location>
<url>https://aclanthology.org/2025.acl-short.86/</url>
</location>
<part>
<date>2025-07</date>
<extent unit="page">
<start>1097</start>
<end>1106</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in SDOH Extraction with Large Language Models
%A Sakib, Fardin Ahsan
%A Zhu, Ziwei
%A Grace, Karen Trister
%A Yetisgen, Meliha
%A Uzuner, Ozlem
%Y Che, Wanxiang
%Y Nabende, Joyce
%Y Shutova, Ekaterina
%Y Pilehvar, Mohammad Taher
%S Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers)
%D 2025
%8 July
%I Association for Computational Linguistics
%C Vienna, Austria
%@ 979-8-89176-252-7
%F sakib-etal-2025-spurious
%X Social determinants of health (SDOH) extraction from clinical text is critical for downstream healthcare analytics. Although large language models (LLMs) have shown promise, they may rely on superficial cues leading to spurious predictions. Using the MIMIC portion of the SHAC (Social History Annotation Corpus) dataset and focusing on drug status extraction as a case study, we demonstrate that mentions of alcohol or smoking can falsely induce models to predict current/past drug use where none is present, while also uncovering concerning gender disparities in model performance. We further evaluate mitigation strategies—such as prompt engineering and chain-of-thought reasoning—to reduce these false positives, providing insights into enhancing LLM reliability in health domains.
%R 10.18653/v1/2025.acl-short.86
%U https://aclanthology.org/2025.acl-short.86/
%U https://doi.org/10.18653/v1/2025.acl-short.86
%P 1097-1106
Markdown (Informal)
[Spurious Correlations and Beyond: Understanding and Mitigating Shortcut Learning in SDOH Extraction with Large Language Models](https://aclanthology.org/2025.acl-short.86/) (Sakib et al., ACL 2025)
ACL