Performance of popular large language models in glaucoma patient education: A randomized controlled study

Purpose: The advent of chatbots based on large language models (LLMs), such as ChatGPT, has significantly transformed knowledge acquisition. However, the application of LLMs in glaucoma patient education remains elusive. In this study, we comprehensively compared the performance of four common LLMs...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuyu Cao, Wei Lu, Runhan Shi, Fuying Liu, Steven Liu, Xinwei Xu, Jin Yang, Guangyu Rong, Changchang Xin, Xujiao Zhou, Xinghuai Sun, Jiaxu Hong
Format: Article
Language:English
Published: Elsevier 2025-05-01
Series:Advances in Ophthalmology Practice and Research
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2667376224000738
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849761717089730560
author Yuyu Cao
Wei Lu
Runhan Shi
Fuying Liu
Steven Liu
Xinwei Xu
Jin Yang
Guangyu Rong
Changchang Xin
Xujiao Zhou
Xinghuai Sun
Jiaxu Hong
author_facet Yuyu Cao
Wei Lu
Runhan Shi
Fuying Liu
Steven Liu
Xinwei Xu
Jin Yang
Guangyu Rong
Changchang Xin
Xujiao Zhou
Xinghuai Sun
Jiaxu Hong
author_sort Yuyu Cao
collection DOAJ
description Purpose: The advent of chatbots based on large language models (LLMs), such as ChatGPT, has significantly transformed knowledge acquisition. However, the application of LLMs in glaucoma patient education remains elusive. In this study, we comprehensively compared the performance of four common LLMs – Qwen, Baichuan 2, ChatGPT-4o, and PaLM 2 – in the context of glaucoma patient education. Methods: Initially, senior ophthalmologists were asked with scoring responses generated by the LLMs, which were answers to the most frequent glaucoma-related questions posed by patients. The Chinese Readability Platform was employed to assess the recommended reading age and reading difficulty score of the four LLMs. Subsequently, optimized models were filtered, and 29 glaucoma patients participated in posing questions to the chatbots and scoring the answers within a real-world clinical setting. Attending ophthalmologists were also required to score the answers across five dimensions: correctness, completeness, readability, helpfulness, and safety. Patients, on the other hand, scored the answers based on three dimensions: satisfaction, readability, and helpfulness. Results: In the first stage, Baichuan 2 and ChatGPT-4o outperformed the other two models, though ChatGPT-4o had higher recommended reading age and reading difficulty scores. In the second stage, both Baichuan 2 and ChatGPT-4o demonstrated exceptional performance among patients and ophthalmologists, with no statistically significant differences observed. Conclusions: Our research identifies Baichuan 2 and ChatGPT-4o as prominent LLMs, offering viable options for glaucoma education.
format Article
id doaj-art-c83d2c79a8ea43deab583aa0b4088353
institution DOAJ
issn 2667-3762
language English
publishDate 2025-05-01
publisher Elsevier
record_format Article
series Advances in Ophthalmology Practice and Research
spelling doaj-art-c83d2c79a8ea43deab583aa0b40883532025-08-20T03:05:56ZengElsevierAdvances in Ophthalmology Practice and Research2667-37622025-05-0152889410.1016/j.aopr.2024.12.002Performance of popular large language models in glaucoma patient education: A randomized controlled studyYuyu Cao0Wei Lu1Runhan Shi2Fuying Liu3Steven Liu4Xinwei Xu5Jin Yang6Guangyu Rong7Changchang Xin8Xujiao Zhou9Xinghuai Sun10Jiaxu Hong11Department of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, ChinaDepartment of Ophthalmology and Vision Science, Shanghai Eye Ear Nose and Throat Hospital, Fudan University, Shanghai, ChinaDepartment of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, ChinaDepartment of Ophthalmology and Vision Science, Shanghai Eye Ear Nose and Throat Hospital, Fudan University, Shanghai, China; People‘s Hospital of Junan, Qingdao University, Shandong, ChinaDepartment of Statistics, College of Liberal Arts & Sciences, University of Illinois Urbana-Champaign, Urbana, Champaign, USAFaculty of Business and Economics, Hong Kong University, Hong Kong, ChinaDepartment of Ophthalmology and Vision Science, Shanghai Eye Ear Nose and Throat Hospital, Fudan University, Shanghai, ChinaDepartment of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, ChinaDepartment of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, ChinaDepartment of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, ChinaDepartment of Ophthalmology and Vision Science, Shanghai Eye Ear Nose and Throat Hospital, Fudan University, Shanghai, China; Corresponding author.Department of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, China; NHC Key Laboratory of Myopia and Related Eye Diseases Shanghai, China; Shanghai Engineering Research Center of Synthetic Immunology, Shanghai, China; Department of Ophthalmology, Children's Hospital of Fudan University, National Pediatric Medical Center of China, Shanghai, China; Corresponding author. Department of Ophthalmology, Eye & ENT Hospital, State Key Laboratory of Medical Neurobiology, Fudan University, Shanghai, 200031, China.Purpose: The advent of chatbots based on large language models (LLMs), such as ChatGPT, has significantly transformed knowledge acquisition. However, the application of LLMs in glaucoma patient education remains elusive. In this study, we comprehensively compared the performance of four common LLMs – Qwen, Baichuan 2, ChatGPT-4o, and PaLM 2 – in the context of glaucoma patient education. Methods: Initially, senior ophthalmologists were asked with scoring responses generated by the LLMs, which were answers to the most frequent glaucoma-related questions posed by patients. The Chinese Readability Platform was employed to assess the recommended reading age and reading difficulty score of the four LLMs. Subsequently, optimized models were filtered, and 29 glaucoma patients participated in posing questions to the chatbots and scoring the answers within a real-world clinical setting. Attending ophthalmologists were also required to score the answers across five dimensions: correctness, completeness, readability, helpfulness, and safety. Patients, on the other hand, scored the answers based on three dimensions: satisfaction, readability, and helpfulness. Results: In the first stage, Baichuan 2 and ChatGPT-4o outperformed the other two models, though ChatGPT-4o had higher recommended reading age and reading difficulty scores. In the second stage, both Baichuan 2 and ChatGPT-4o demonstrated exceptional performance among patients and ophthalmologists, with no statistically significant differences observed. Conclusions: Our research identifies Baichuan 2 and ChatGPT-4o as prominent LLMs, offering viable options for glaucoma education.http://www.sciencedirect.com/science/article/pii/S2667376224000738GlaucomaPatient educationLarge language modelsChatGPTPaLMBaichuan
spellingShingle Yuyu Cao
Wei Lu
Runhan Shi
Fuying Liu
Steven Liu
Xinwei Xu
Jin Yang
Guangyu Rong
Changchang Xin
Xujiao Zhou
Xinghuai Sun
Jiaxu Hong
Performance of popular large language models in glaucoma patient education: A randomized controlled study
Advances in Ophthalmology Practice and Research
Glaucoma
Patient education
Large language models
ChatGPT
PaLM
Baichuan
title Performance of popular large language models in glaucoma patient education: A randomized controlled study
title_full Performance of popular large language models in glaucoma patient education: A randomized controlled study
title_fullStr Performance of popular large language models in glaucoma patient education: A randomized controlled study
title_full_unstemmed Performance of popular large language models in glaucoma patient education: A randomized controlled study
title_short Performance of popular large language models in glaucoma patient education: A randomized controlled study
title_sort performance of popular large language models in glaucoma patient education a randomized controlled study
topic Glaucoma
Patient education
Large language models
ChatGPT
PaLM
Baichuan
url http://www.sciencedirect.com/science/article/pii/S2667376224000738
work_keys_str_mv AT yuyucao performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT weilu performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT runhanshi performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT fuyingliu performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT stevenliu performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT xinweixu performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT jinyang performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT guangyurong performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT changchangxin performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT xujiaozhou performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT xinghuaisun performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy
AT jiaxuhong performanceofpopularlargelanguagemodelsinglaucomapatienteducationarandomizedcontrolledstudy