@inproceedings{ibrahim-etal-2025-bridging,
title = "Bridging Dialectal Gaps in {A}rabic Medical {LLM}s through Model Merging",
author = "Ibrahim, Ahmed and
Hosseini, Abdullah and
Helmy, Hoda and
Lakhdhar, Wafa and
Serag, Ahmed",
editor = "Darwish, Kareem and
Ali, Ahmed and
Abu Farha, Ibrahim and
Touileb, Samia and
Zitouni, Imed and
Abdelali, Ahmed and
Al-Ghamdi, Sharefah and
Alkhereyf, Sakhar and
Zaghouani, Wajdi and
Khalifa, Salam and
AlKhamissi, Badr and
Almatham, Rawan and
Hamed, Injy and
Alyafeai, Zaid and
Alowisheq, Areeb and
Inoue, Go and
Mrini, Khalil and
Alshammari, Waad",
booktitle = "Proceedings of The Third Arabic Natural Language Processing Conference",
month = nov,
year = "2025",
address = "Suzhou, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.arabicnlp-main.27/",
pages = "338--346",
ISBN = "979-8-89176-352-4",
abstract = "The linguistic fragmentation of Arabic, with over 30 dialects exhibiting low mutual intelligibility, presents a critical challenge for deploying natural language processing (NLP) in healthcare. Conventional fine-tuning of large language models (LLMs) for each dialect is computationally prohibitive and operationally unsustainable. In this study, we explore model merging as a scalable alternative by integrating three pre-trained LLMs{---}a medical domain expert, an Egyptian Arabic model, and a Moroccan Darija model{---}into a unified system without additional fine-tuning. We introduce a novel evaluation framework that assesses both dialectal fidelity via dual evaluation: LLM-based automated scoring and human assessments by native speakers. Our results demonstrate that the merged model effectively handles cross-dialect medical scenarios, such as interpreting Moroccan Darija inputs for Egyptian Arabic-speaking clinicians, while maintaining high clinical relevance. The merging process reduced computational cost by over 60{\%} compared to per-dialect fine-tuning, highlighting its viability for resource-constrained settings. This work offers a promising path for building dialect-aware medical LLMs at scale, with implications for broader deployment across linguistically diverse regions."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ibrahim-etal-2025-bridging">
<titleInfo>
<title>Bridging Dialectal Gaps in Arabic Medical LLMs through Model Merging</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ahmed</namePart>
<namePart type="family">Ibrahim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Abdullah</namePart>
<namePart type="family">Hosseini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hoda</namePart>
<namePart type="family">Helmy</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wafa</namePart>
<namePart type="family">Lakhdhar</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ahmed</namePart>
<namePart type="family">Serag</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>Proceedings of The Third Arabic Natural Language Processing Conference</title>
</titleInfo>
<name type="personal">
<namePart type="given">Kareem</namePart>
<namePart type="family">Darwish</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ahmed</namePart>
<namePart type="family">Ali</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ibrahim</namePart>
<namePart type="family">Abu Farha</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Samia</namePart>
<namePart type="family">Touileb</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Imed</namePart>
<namePart type="family">Zitouni</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ahmed</namePart>
<namePart type="family">Abdelali</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sharefah</namePart>
<namePart type="family">Al-Ghamdi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sakhar</namePart>
<namePart type="family">Alkhereyf</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wajdi</namePart>
<namePart type="family">Zaghouani</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Salam</namePart>
<namePart type="family">Khalifa</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Badr</namePart>
<namePart type="family">AlKhamissi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rawan</namePart>
<namePart type="family">Almatham</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Injy</namePart>
<namePart type="family">Hamed</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zaid</namePart>
<namePart type="family">Alyafeai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Areeb</namePart>
<namePart type="family">Alowisheq</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Go</namePart>
<namePart type="family">Inoue</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Khalil</namePart>
<namePart type="family">Mrini</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Waad</namePart>
<namePart type="family">Alshammari</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">979-8-89176-352-4</identifier>
</relatedItem>
<abstract>The linguistic fragmentation of Arabic, with over 30 dialects exhibiting low mutual intelligibility, presents a critical challenge for deploying natural language processing (NLP) in healthcare. Conventional fine-tuning of large language models (LLMs) for each dialect is computationally prohibitive and operationally unsustainable. In this study, we explore model merging as a scalable alternative by integrating three pre-trained LLMs—a medical domain expert, an Egyptian Arabic model, and a Moroccan Darija model—into a unified system without additional fine-tuning. We introduce a novel evaluation framework that assesses both dialectal fidelity via dual evaluation: LLM-based automated scoring and human assessments by native speakers. Our results demonstrate that the merged model effectively handles cross-dialect medical scenarios, such as interpreting Moroccan Darija inputs for Egyptian Arabic-speaking clinicians, while maintaining high clinical relevance. The merging process reduced computational cost by over 60% compared to per-dialect fine-tuning, highlighting its viability for resource-constrained settings. This work offers a promising path for building dialect-aware medical LLMs at scale, with implications for broader deployment across linguistically diverse regions.</abstract>
<identifier type="citekey">ibrahim-etal-2025-bridging</identifier>
<location>
<url>https://aclanthology.org/2025.arabicnlp-main.27/</url>
</location>
<part>
<date>2025-11</date>
<extent unit="page">
<start>338</start>
<end>346</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Bridging Dialectal Gaps in Arabic Medical LLMs through Model Merging
%A Ibrahim, Ahmed
%A Hosseini, Abdullah
%A Helmy, Hoda
%A Lakhdhar, Wafa
%A Serag, Ahmed
%Y Darwish, Kareem
%Y Ali, Ahmed
%Y Abu Farha, Ibrahim
%Y Touileb, Samia
%Y Zitouni, Imed
%Y Abdelali, Ahmed
%Y Al-Ghamdi, Sharefah
%Y Alkhereyf, Sakhar
%Y Zaghouani, Wajdi
%Y Khalifa, Salam
%Y AlKhamissi, Badr
%Y Almatham, Rawan
%Y Hamed, Injy
%Y Alyafeai, Zaid
%Y Alowisheq, Areeb
%Y Inoue, Go
%Y Mrini, Khalil
%Y Alshammari, Waad
%S Proceedings of The Third Arabic Natural Language Processing Conference
%D 2025
%8 November
%I Association for Computational Linguistics
%C Suzhou, China
%@ 979-8-89176-352-4
%F ibrahim-etal-2025-bridging
%X The linguistic fragmentation of Arabic, with over 30 dialects exhibiting low mutual intelligibility, presents a critical challenge for deploying natural language processing (NLP) in healthcare. Conventional fine-tuning of large language models (LLMs) for each dialect is computationally prohibitive and operationally unsustainable. In this study, we explore model merging as a scalable alternative by integrating three pre-trained LLMs—a medical domain expert, an Egyptian Arabic model, and a Moroccan Darija model—into a unified system without additional fine-tuning. We introduce a novel evaluation framework that assesses both dialectal fidelity via dual evaluation: LLM-based automated scoring and human assessments by native speakers. Our results demonstrate that the merged model effectively handles cross-dialect medical scenarios, such as interpreting Moroccan Darija inputs for Egyptian Arabic-speaking clinicians, while maintaining high clinical relevance. The merging process reduced computational cost by over 60% compared to per-dialect fine-tuning, highlighting its viability for resource-constrained settings. This work offers a promising path for building dialect-aware medical LLMs at scale, with implications for broader deployment across linguistically diverse regions.
%U https://aclanthology.org/2025.arabicnlp-main.27/
%P 338-346
Markdown (Informal)
[Bridging Dialectal Gaps in Arabic Medical LLMs through Model Merging](https://aclanthology.org/2025.arabicnlp-main.27/) (Ibrahim et al., ArabicNLP 2025)
ACL