@inproceedings{van-veen-etal-2023-radadapt,
title = "{R}ad{A}dapt: Radiology Report Summarization via Lightweight Domain Adaptation of Large Language Models",
author = "Van Veen, Dave and
Van Uden, Cara and
Attias, Maayane and
Pareek, Anuj and
Bluethgen, Christian and
Polacin, Malgorzata and
Chiu, Wah and
Delbrouck, Jean-Benoit and
Zambrano Chaves, Juan and
Langlotz, Curtis and
Chaudhari, Akshay and
Pauly, John",
editor = "Demner-fushman, Dina and
Ananiadou, Sophia and
Cohen, Kevin",
booktitle = "The 22nd Workshop on Biomedical Natural Language Processing and BioNLP Shared Tasks",
month = jul,
year = "2023",
address = "Toronto, Canada",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2023.bionlp-1.42",
doi = "10.18653/v1/2023.bionlp-1.42",
pages = "449--460",
abstract = "We systematically investigate lightweight strategies to adapt large language models (LLMs) for the task of radiology report summarization (RRS). Specifically, we focus on domain adaptation via pretraining (on natural language, biomedical text, or clinical text) and via discrete prompting or parameter-efficient fine-tuning. Our results consistently achieve best performance by maximally adapting to the task via pretraining on clinical text and fine-tuning on RRS examples. Importantly, this method fine-tunes a mere 0.32{\%} of parameters throughout the model, in contrast to end-to-end fine-tuning (100{\%} of parameters). Additionally, we study the effect of in-context examples and out-of-distribution (OOD) training before concluding with a radiologist reader study and qualitative analysis. Our findings highlight the importance of domain adaptation in RRS and provide valuable insights toward developing effective natural language processing solutions for clinical tasks.",
}
<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="van-veen-etal-2023-radadapt">
<titleInfo>
<title>RadAdapt: Radiology Report Summarization via Lightweight Domain Adaptation of Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dave</namePart>
<namePart type="family">Van Veen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Cara</namePart>
<namePart type="family">Van Uden</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Maayane</namePart>
<namePart type="family">Attias</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Anuj</namePart>
<namePart type="family">Pareek</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Christian</namePart>
<namePart type="family">Bluethgen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Malgorzata</namePart>
<namePart type="family">Polacin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wah</namePart>
<namePart type="family">Chiu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jean-Benoit</namePart>
<namePart type="family">Delbrouck</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Juan</namePart>
<namePart type="family">Zambrano Chaves</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Curtis</namePart>
<namePart type="family">Langlotz</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Akshay</namePart>
<namePart type="family">Chaudhari</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">John</namePart>
<namePart type="family">Pauly</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2023-07</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>The 22nd Workshop on Biomedical Natural Language Processing and BioNLP Shared Tasks</title>
</titleInfo>
<name type="personal">
<namePart type="given">Dina</namePart>
<namePart type="family">Demner-fushman</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sophia</namePart>
<namePart type="family">Ananiadou</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Kevin</namePart>
<namePart type="family">Cohen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Toronto, Canada</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>We systematically investigate lightweight strategies to adapt large language models (LLMs) for the task of radiology report summarization (RRS). Specifically, we focus on domain adaptation via pretraining (on natural language, biomedical text, or clinical text) and via discrete prompting or parameter-efficient fine-tuning. Our results consistently achieve best performance by maximally adapting to the task via pretraining on clinical text and fine-tuning on RRS examples. Importantly, this method fine-tunes a mere 0.32% of parameters throughout the model, in contrast to end-to-end fine-tuning (100% of parameters). Additionally, we study the effect of in-context examples and out-of-distribution (OOD) training before concluding with a radiologist reader study and qualitative analysis. Our findings highlight the importance of domain adaptation in RRS and provide valuable insights toward developing effective natural language processing solutions for clinical tasks.</abstract>
<identifier type="citekey">van-veen-etal-2023-radadapt</identifier>
<identifier type="doi">10.18653/v1/2023.bionlp-1.42</identifier>
<location>
<url>https://aclanthology.org/2023.bionlp-1.42</url>
</location>
<part>
<date>2023-07</date>
<extent unit="page">
<start>449</start>
<end>460</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T RadAdapt: Radiology Report Summarization via Lightweight Domain Adaptation of Large Language Models
%A Van Veen, Dave
%A Van Uden, Cara
%A Attias, Maayane
%A Pareek, Anuj
%A Bluethgen, Christian
%A Polacin, Malgorzata
%A Chiu, Wah
%A Delbrouck, Jean-Benoit
%A Zambrano Chaves, Juan
%A Langlotz, Curtis
%A Chaudhari, Akshay
%A Pauly, John
%Y Demner-fushman, Dina
%Y Ananiadou, Sophia
%Y Cohen, Kevin
%S The 22nd Workshop on Biomedical Natural Language Processing and BioNLP Shared Tasks
%D 2023
%8 July
%I Association for Computational Linguistics
%C Toronto, Canada
%F van-veen-etal-2023-radadapt
%X We systematically investigate lightweight strategies to adapt large language models (LLMs) for the task of radiology report summarization (RRS). Specifically, we focus on domain adaptation via pretraining (on natural language, biomedical text, or clinical text) and via discrete prompting or parameter-efficient fine-tuning. Our results consistently achieve best performance by maximally adapting to the task via pretraining on clinical text and fine-tuning on RRS examples. Importantly, this method fine-tunes a mere 0.32% of parameters throughout the model, in contrast to end-to-end fine-tuning (100% of parameters). Additionally, we study the effect of in-context examples and out-of-distribution (OOD) training before concluding with a radiologist reader study and qualitative analysis. Our findings highlight the importance of domain adaptation in RRS and provide valuable insights toward developing effective natural language processing solutions for clinical tasks.
%R 10.18653/v1/2023.bionlp-1.42
%U https://aclanthology.org/2023.bionlp-1.42
%U https://doi.org/10.18653/v1/2023.bionlp-1.42
%P 449-460
Markdown (Informal)
[RadAdapt: Radiology Report Summarization via Lightweight Domain Adaptation of Large Language Models](https://aclanthology.org/2023.bionlp-1.42) (Van Veen et al., BioNLP 2023)
ACL
- Dave Van Veen, Cara Van Uden, Maayane Attias, Anuj Pareek, Christian Bluethgen, Malgorzata Polacin, Wah Chiu, Jean-Benoit Delbrouck, Juan Zambrano Chaves, Curtis Langlotz, Akshay Chaudhari, and John Pauly. 2023. RadAdapt: Radiology Report Summarization via Lightweight Domain Adaptation of Large Language Models. In The 22nd Workshop on Biomedical Natural Language Processing and BioNLP Shared Tasks, pages 449–460, Toronto, Canada. Association for Computational Linguistics.