@inproceedings{wu-etal-2025-assessing,
title = "Assessing and Mitigating Medical Knowledge Drift and Conflicts in Large Language Models",
author = "Wu, Weiyi and
Xu, Xinwen and
Gao, Chongyang and
Diao, Xingjian and
Li, Siting and
Salas, Lucas A. and
Gui, Jiang",
editor = "Christodoulopoulos, Christos and
Chakraborty, Tanmoy and
Rose, Carolyn and
Peng, Violet",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.findings-emnlp.38/",
pages = "707--730",
ISBN = "979-8-89176-335-7",
abstract = "Large Language Models (LLMs) offer transformative potential across diverse fields, yet their safe and effective deployment is hindered by inherent knowledge conflicts{---}stemming from temporal evolution, divergent sources, and contradictory guidelines. This challenge is particularly acute in medicine, an interdisciplinary frontier for NLP. Rapid medical concept drift can lead LLMs to provide incorrect or outdated advice, impacting their utility and the broader societal benefits of NLP advances. This study introduces ConflictMedQA, a benchmark designed to systematically evaluate how LLMs manage varied knowledge conflicts in clinical guidelines. Our assessment of seven state-of-the-art models across 4,290 scenarios reveals significant difficulties in rejecting incorrect recommendations and frequent endorsement of conflicting advice, highlighting an important gap for NLP systems intended for real-world impact. We explore two fundamental mitigation approaches: retrieval-augmented generation and preference fine-tuning via direct preference optimization. While each offers improvements, their synergistic combination yields the best results. These findings emphasize the need for LLMs to discern subtle but critical guideline conflicts. This is a crucial step in advancing NLP{'}s capabilities and ensuring its dependable application in critical societal domains. The proposed dataset is available at https://huggingface.co/datasets/RDBH/DriftMed."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="wu-etal-2025-assessing">
<titleInfo>
<title>Assessing and Mitigating Medical Knowledge Drift and Conflicts in Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Weiyi</namePart>
<namePart type="family">Wu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xinwen</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chongyang</namePart>
<namePart type="family">Gao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xingjian</namePart>
<namePart type="family">Diao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siting</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lucas</namePart>
<namePart type="given">A</namePart>
<namePart type="family">Salas</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jiang</namePart>
<namePart type="family">Gui</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2025-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2025</title>
</titleInfo>
<name type="personal">
<namePart type="given">Christos</namePart>
<namePart type="family">Christodoulopoulos</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Carolyn</namePart>
<namePart type="family">Rose</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Violet</namePart>
<namePart type="family">Peng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Suzhou, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-335-7</identifier>
</relatedItem>
<abstract>Large Language Models (LLMs) offer transformative potential across diverse fields, yet their safe and effective deployment is hindered by inherent knowledge conflicts—stemming from temporal evolution, divergent sources, and contradictory guidelines. This challenge is particularly acute in medicine, an interdisciplinary frontier for NLP. Rapid medical concept drift can lead LLMs to provide incorrect or outdated advice, impacting their utility and the broader societal benefits of NLP advances. This study introduces ConflictMedQA, a benchmark designed to systematically evaluate how LLMs manage varied knowledge conflicts in clinical guidelines. Our assessment of seven state-of-the-art models across 4,290 scenarios reveals significant difficulties in rejecting incorrect recommendations and frequent endorsement of conflicting advice, highlighting an important gap for NLP systems intended for real-world impact. We explore two fundamental mitigation approaches: retrieval-augmented generation and preference fine-tuning via direct preference optimization. While each offers improvements, their synergistic combination yields the best results. These findings emphasize the need for LLMs to discern subtle but critical guideline conflicts. This is a crucial step in advancing NLP’s capabilities and ensuring its dependable application in critical societal domains. The proposed dataset is available at https://huggingface.co/datasets/RDBH/DriftMed.</abstract>
<identifier type="citekey">wu-etal-2025-assessing</identifier>
<location>
<url>https://aclanthology.org/2025.findings-emnlp.38/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>707</start>
<end>730</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Assessing and Mitigating Medical Knowledge Drift and Conflicts in Large Language Models
%A Wu, Weiyi
%A Xu, Xinwen
%A Gao, Chongyang
%A Diao, Xingjian
%A Li, Siting
%A Salas, Lucas A.
%A Gui, Jiang
%Y Christodoulopoulos, Christos
%Y Chakraborty, Tanmoy
%Y Rose, Carolyn
%Y Peng, Violet
%S Findings of the Association for Computational Linguistics: EMNLP 2025
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-335-7
%F wu-etal-2025-assessing
%X Large Language Models (LLMs) offer transformative potential across diverse fields, yet their safe and effective deployment is hindered by inherent knowledge conflicts—stemming from temporal evolution, divergent sources, and contradictory guidelines. This challenge is particularly acute in medicine, an interdisciplinary frontier for NLP. Rapid medical concept drift can lead LLMs to provide incorrect or outdated advice, impacting their utility and the broader societal benefits of NLP advances. This study introduces ConflictMedQA, a benchmark designed to systematically evaluate how LLMs manage varied knowledge conflicts in clinical guidelines. Our assessment of seven state-of-the-art models across 4,290 scenarios reveals significant difficulties in rejecting incorrect recommendations and frequent endorsement of conflicting advice, highlighting an important gap for NLP systems intended for real-world impact. We explore two fundamental mitigation approaches: retrieval-augmented generation and preference fine-tuning via direct preference optimization. While each offers improvements, their synergistic combination yields the best results. These findings emphasize the need for LLMs to discern subtle but critical guideline conflicts. This is a crucial step in advancing NLP’s capabilities and ensuring its dependable application in critical societal domains. The proposed dataset is available at https://huggingface.co/datasets/RDBH/DriftMed.
%U https://aclanthology.org/2025.findings-emnlp.38/
%P 707-730
Markdown (Informal)
[Assessing and Mitigating Medical Knowledge Drift and Conflicts in Large Language Models](https://aclanthology.org/2025.findings-emnlp.38/) (Wu et al., Findings 2025)
ACL