The growing presence of AI-generated content in ophthalmology: a retrospective bibliographic analysis.

IF 2.8 4区医学 Q1 OPHTHALMOLOGY

Canadian journal of ophthalmology. Journal canadien d'ophtalmologie Pub Date : 2025-10-09 DOI:10.1016/j.jcjo.2025.09.004

Michael Balas, Xiaole Li, Patrick Ji, Parnian Arjmand

{"title":"The growing presence of AI-generated content in ophthalmology: a retrospective bibliographic analysis.","authors":"Michael Balas, Xiaole Li, Patrick Ji, Parnian Arjmand","doi":"10.1016/j.jcjo.2025.09.004","DOIUrl":null,"url":null,"abstract":"Objective: To quantify the prevalence and trends of artificial intelligence (AI)-generated content in ophthalmology manuscripts, particularly following the public release of OpenAI's ChatGPT on November 30, 2022.Methods: A retrospective bibliographic analysis was conducted on 1 036 manuscripts from 30 ophthalmology journals, divided into pre-December 2022 (519 manuscripts) and post-December 2022 (517 manuscripts) periods. AI-generated content was evaluated using the Originality Standard 2.0.0 model, which calculates AI probability scores (AIPS) ranging from 0% to 100%. Readability metrics (e.g., Flesch-Kincaid Score) and journal impact metrics (e.g., impact factor) were analyzed.Results: AIPS remained stable from 2014 to 2022 but increased significantly after December 2022 (p < 0.001). The mean AIPS rose from 4.95% in 2022 to 11.2% by mid-2024, with projections estimating 17.51% by mid-2026. Editorials exhibited the highest mean AIPS (12.8%), while surgical technique studies had the lowest (4.33%). Higher AIPS were associated with lower journal impact factors (Spearman's ρ = -0.54; p < 0.001) and simpler language, as reflected by lower Automated Readability Index scores (Spearman's ρ = -0.12; p < 0.005). None of the included manuscripts disclosed AI usage, including 44 manuscripts with AIPS exceeding 25%.Conclusions: AI-generated content in ophthalmology has risen significantly since ChatGPT's release. Higher AIPS correlates with lower journal impact factors and reduced literary complexity. The lack of AI usage disclosure raises ethical concerns and emphasizes the need for transparent reporting and guidelines to ensure the integrity of scientific research.","PeriodicalId":9606,"journal":{"name":"Canadian journal of ophthalmology. Journal canadien d'ophtalmologie","volume":" ","pages":""},"PeriodicalIF":2.8000,"publicationDate":"2025-10-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Canadian journal of ophthalmology. Journal canadien d'ophtalmologie","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1016/j.jcjo.2025.09.004","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"OPHTHALMOLOGY","Score":null,"Total":0}

引用次数: 0

Abstract

Objective: To quantify the prevalence and trends of artificial intelligence (AI)-generated content in ophthalmology manuscripts, particularly following the public release of OpenAI's ChatGPT on November 30, 2022.

Methods: A retrospective bibliographic analysis was conducted on 1 036 manuscripts from 30 ophthalmology journals, divided into pre-December 2022 (519 manuscripts) and post-December 2022 (517 manuscripts) periods. AI-generated content was evaluated using the Originality Standard 2.0.0 model, which calculates AI probability scores (AIPS) ranging from 0% to 100%. Readability metrics (e.g., Flesch-Kincaid Score) and journal impact metrics (e.g., impact factor) were analyzed.

Results: AIPS remained stable from 2014 to 2022 but increased significantly after December 2022 (p < 0.001). The mean AIPS rose from 4.95% in 2022 to 11.2% by mid-2024, with projections estimating 17.51% by mid-2026. Editorials exhibited the highest mean AIPS (12.8%), while surgical technique studies had the lowest (4.33%). Higher AIPS were associated with lower journal impact factors (Spearman's ρ = -0.54; p < 0.001) and simpler language, as reflected by lower Automated Readability Index scores (Spearman's ρ = -0.12; p < 0.005). None of the included manuscripts disclosed AI usage, including 44 manuscripts with AIPS exceeding 25%.

Conclusions: AI-generated content in ophthalmology has risen significantly since ChatGPT's release. Higher AIPS correlates with lower journal impact factors and reduced literary complexity. The lack of AI usage disclosure raises ethical concerns and emphasizes the need for transparent reporting and guidelines to ensure the integrity of scientific research.

查看原文本刊更多论文

人工智能生成的内容在眼科中的日益增长：回顾性书目分析。

背景：量化眼科手稿中人工智能（AI）生成内容的流行程度和趋势，特别是在2022年11月30日OpenAI的ChatGPT公开发布之后。方法：对30篇眼科期刊1 036篇文献进行回顾性文献分析，分为2022年12月前（519篇）和2022年12月后（517篇）两个时间段。AI生成的内容使用独创性标准2.0.0模型进行评估，该模型计算AI概率分数（AIPS），范围从0%到100%。分析了可读性指标（如Flesch-Kincaid评分）和期刊影响指标（如影响因子）。结果：AIPS在2014 - 2022年保持稳定，但在2022年12月之后显著上升（p < 0.001）。平均AIPS从2022年的4.95%上升到2024年中期的11.2%，预计到2026年中期将达到17.51%。社论的平均AIPS最高（12.8%），外科技术研究的平均AIPS最低（4.33%）。较高的AIPS与较低的期刊影响因子（Spearman的ρ = -0.54;p < 0.001）和较简单的语言相关，反映在较低的自动可读性指数得分上（Spearman的ρ = -0.12;p < 0.005）。纳入的稿件均未披露AI使用情况，其中AIPS超过25%的稿件有44篇。结论：自ChatGPT发布以来，人工智能生成的眼科内容显著增加。较高的AIPS与较低的期刊影响因子和较低的文献复杂性相关。缺乏对人工智能使用情况的披露引发了伦理问题，并强调需要透明的报告和指导方针，以确保科学研究的完整性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Canadian journal of ophthalmology. Journal canadien d'ophtalmologie 医学-眼科学

CiteScore

3.20

自引率

4.80%

发文量

223

审稿时长

38 days

期刊介绍： Official journal of the Canadian Ophthalmological Society. The Canadian Journal of Ophthalmology (CJO) is the official journal of the Canadian Ophthalmological Society and is committed to timely publication of original, peer-reviewed ophthalmology and vision science articles.