文献速递技术与安全

影像学期刊关于稿件生成中使用人工智能政策的比较综述

A Comparative Review of Imaging Journal Policies for Use of AI in Manuscript Generation.
PubMed 文献 · Acad Radiol · 2024-05-20

该横断面研究于2023年10月13日至16日评估了112种MEDLINE收录的影像学期刊的作者指南,其中69种(61.6%)有AI使用政策,40种(57.9%)提及AI生成图片的具体政策;有AI政策的期刊在CiteScore、Scientific Journal Ranking和Journal Citation Indicator上显著更高。

要点

  • 112种影像学期刊中69种(61.6%)有AI使用政策,40种(57.9%)提及AI生成图片政策。
  • 有AI政策期刊的CiteScore为4.9 vs 无政策期刊4,P=0.023。
  • 有AI政策期刊的Scientific Journal Ranking为0.75 vs 无政策期刊0.54,P=0.010。

核心结论

  • 多数影像学期刊提供了AI生成内容的指南,但仍有相当数量的期刊没有AI使用政策或不要求披露非人类创作的稿件;有明确AI政策的期刊具有更高的引用和影响力评分。

研究设计与样本

  • 横断面研究,纳入112种MEDLINE收录的影像学期刊,于2023年10月13日至16日评估作者指南和编辑政策;期刊根据主题和与放射学会的关联识别;使用Wilcoxon符号秩检验比较有和无AI使用政策期刊的科学计量学评分。

诊断标准与关键数值

  • 112种影像学期刊中69种(61.6%)有AI使用政策,40种(57.9%)提及AI生成图片的具体政策。
  • 有AI政策期刊的CiteScore为4.9,无AI政策期刊为4,P=0.023。
  • 有AI政策期刊的Source Normalized Impact per Paper为1.12,无AI政策期刊为0.83,P=0.06。
  • 有AI政策期刊的Scientific Journal Ranking为0.75,无AI政策期刊为0.54,P=0.010。
  • 有AI政策期刊的Journal Citation Indicator为0.77,无AI政策期刊为0.62,P=0.038。

局限性

  • 原文未报告局限性。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

英文原文(摘要)· 点击展开

RATIONALE AND OBJECTIVES: Artificial intelligence (AI) technologies are rapidly evolving and offering new advances almost on a day-by-day basis, including various tools for manuscript generation and modification. On the other hand, these potentially time- and effort-saving solutions come with potential bias, factual error, and plagiarism risks. Some journals have started to update their author guidelines in reference to AI-generated or AI-assisted manuscripts. The purpose of this paper is to evaluate author guidelines for including AI use policies in radiology journals and compare scientometric data between journals with and without explicit AI use policies.

MATERIALS AND METHODS: This cross-sectional study included 112 MEDLINE-indexed imaging journals and evaluated their author guidelines between 13 October 2023 and 16 October 2023. Journals were identified based on subject matter and association with a radiological society. The authors' guidelines and editorial policies were evaluated for the use of AI in manuscript preparation and specific AI-generated image policies. We assessed the existence of an AI usage policy among subspecialty imaging journals. The scientometric scores of journals with and without AI use policies were compared using the Wilcoxon signed-rank test.

RESULTS: Among 112 MEDLINE-indexed radiology journals, 80 journals were affiliated with an imaging society, and 32 were not. 69 (61.6%) of 112 imaging journals had an AI usage policy, and 40 (57.9%) of 69 mentioned a specific policy about AI-generated figures. CiteScore (4.9 vs 4, p = 0.023), Source Normalized Impact per Paper (1.12 vs 0.83, p = 0.06), Scientific Journal Ranking (0.75 vs 0.54, p = 0.010) and Journal Citation Indicator (0.77 vs 0.62, p = 0.038) were significantly higher in journals with an AI policy.

CONCLUSION: The majority of imaging journals provide guidelines for AI-generated content, but still, a substantial number of journals do not have AI usage policies or do not require disclosure for non-human-created manuscripts. Journals with an established AI policy had higher citation and impact scores.

来源:Acad Radiol,查看原文。版权归原作者所有。

引用

  1. Simsek O, Manteghinejad A, Vossough A, et al. A Comparative Review of Imaging Journal Policies for Use of AI in Manuscript Generation. Acad Radiol. 2024. PMID: 38772797. doi:10.1016/j.acra.2024.05.006

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。