Academic Journal
Προβολή σε EDS

Evaluating the Feasibility of Artificial Intelligence in Generating Visual Abstracts: A Pilot Study.

Λεπτομέρειες βιβλιογραφικής εγγραφής
Τίτλος: Evaluating the Feasibility of Artificial Intelligence in Generating Visual Abstracts: A Pilot Study.
Συγγραφείς: Akhund R; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama., Mainali B; Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama., Izhar A; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama., Muñoz Wilson DD; Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama., Gillis A; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama., Fazendin JM; Department of Surgery, University of Oklahoma, Oklahoma City, Oklahoma., Chen H; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama., Lindeman B; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama. Electronic address: blindeman@uabmc.edu.
Πηγή: The Journal of surgical research [J Surg Res] 2026 Oct; Vol. 326, pp. 525-543. Date of Electronic Publication: 2026 Aug 05.
Τύπος έκδοσης: Journal Article; Evaluation Study
Γλώσσα: English
Στοιχεία περιοδικού: Publisher: Academic Press Country of Publication: United States NLM ID: 0376340 Publication Model: Print-Electronic Cited Medium: Internet ISSN: 1095-8673 (Electronic) Linking ISSN: 00224804 NLM ISO Abbreviation: J Surg Res Subsets: MEDLINE
Imprint Name(s): Publication: New York, NY : Academic Press
Original Publication: Philadelphia [etc.]
Ιατρικοί όροι (MeSH): Abstracting and Indexing*/methods , Artificial Intelligence*, Pilot Projects ; Feasibility Studies ; Generative Artificial Intelligence ; Humans ; Large Language Models
Περίληψη: Introduction: Visual abstracts (VAs) have emerged as an efficient tool to disseminate scientific research, offering faster processing and improved information retention compared to traditional text-only abstracts. Concurrently, artificial intelligence (AI) has seen rapid advancements in language models and image generation. This study aims to evaluate the feasibility of using GPT-4o, and its more advanced model GPT-5.3, available on GPT-Pro paid version to generate VAs for scientific literature.
Methods: A pilot study was conducted using a dataset comprised of 26 pairs of visual and textual abstracts. The dataset was split into a training set (n = 20, 77%) and a testing set (n = 6, 23%) to approximate the commonly used 80-20 rule in machine learning while accommodating our sample size. GPT4o and GPT 5.3 were used as the AI models. The system was fine-tuned on the training set, with each input including the textual abstract and corresponding VA. In the testing phase, GPT4o and GPT 5.3 were provided the six-test set textual abstracts and a template generated from expert guidelines, prompting it to generate VAs. GPT 5.3 output was further refined with two iterative prompts after its first generated image. Our research team then compared the AI-generated VAs to the journal-created VAs, focusing on adherence to the template, accuracy of information representation, visual clarity, and overall design consistency.
Results: VAs produced by GPT4o demonstrated significant limitations. They consistently deviated from the provided template, exhibiting excessive imagery and text that often obscured the main message. All six generated VAs contained on average 17 instances (range 12-22) of misspelled and distorted text, impeding readability and comprehension. Notably, there was a lack of consistent structure across the AI-generated VAs, with each output varying considerably in format and style. Information accuracy was also compromised, with the AI output misrepresenting or omitting key data points from the original abstracts for the entire test cohort. The visual clarity was suboptimal, with cluttered layouts and poor color schemes that failed to effectively highlight the most important information. In contrast, VAs generated by GPT 5.3, before and after the iterative prompts, improved widely upon all four domains. While it still made errors regarding journal logo and author name, the rest of the design and consistency was comparable to the human-generated VAs.
Conclusions: Our pilot study demonstrates that while the GPT4o model lacks the capability to produce VAs suitable for scientific knowledge dissemination, a stark improvement was noticed with the most recent GPT-5.3 model, with its ability to adhere to the standardized template and produce images with clarity, accuracy, and visual consistency, required for effective research communication, rendering it useful for generation of a first draft of VAs.
(Copyright © 2026 Elsevier Inc. All rights reserved.)
Contributed Indexing: Keywords: Artificial intelligence; Image generation; Visual abstract
Entry Date(s): Date Created: 20260805 Date Completed: 20261008 Latest Revision: 20261008
Update Code: 20261008
DOI: 10.1016/j.jss.2026.05.087
PMID: 42556195
Βάση Δεδομένων: MEDLINE
FullText Text:
  Availability: 0
CustomLinks:
  – Url: https://www.doi.org/10.1016/j.jss.2026.05.087?
    Name: ScienceDirect (all content) (s7799221)
    Category: fullText
    Text: View record from ScienceDirect
    MouseOverText: View record from ScienceDirect
  – Url: https://resolver.ebsco.com/c/fiv2js/result?sid=EBSCO:cmedm&genre=article&issn=10958673&ISBN=&volume=326&issue=&date=20261001&spage=525&pages=525-543&title=The Journal of surgical research&atitle=Evaluating%20the%20Feasibility%20of%20Artificial%20Intelligence%20in%20Generating%20Visual%20Abstracts%3A%20A%20Pilot%20Study.&aulast=Akhund%20R&id=DOI:10.1016/j.jss.2026.05.087
    Name: Full Text Finder (for New FTF UI) (ns324271)
    Category: fullText
    Text: Full Text Finder
    MouseOverText: Full Text Finder
Header DbId: cmedm
DbLabel: MEDLINE
An: 42556195
AccessLevel: 3
PubType: Academic Journal
PubTypeId: academicJournal
PreciseRelevancyScore: 0
IllustrationInfo
Items – Name: Title
  Label: Title
  Group: Ti
  Data: Evaluating the Feasibility of Artificial Intelligence in Generating Visual Abstracts: A Pilot Study.
– Name: Author
  Label: Authors
  Group: Au
  Data: <searchLink fieldCode="AU" term="%22Akhund+R%22">Akhund R</searchLink>; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Mainali+B%22">Mainali B</searchLink>; Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Izhar+A%22">Izhar A</searchLink>; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Muñoz+Wilson+DD%22">Muñoz Wilson DD</searchLink>; Heersink School of Medicine, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Gillis+A%22">Gillis A</searchLink>; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Fazendin+JM%22">Fazendin JM</searchLink>; Department of Surgery, University of Oklahoma, Oklahoma City, Oklahoma.<br /><searchLink fieldCode="AU" term="%22Chen+H%22">Chen H</searchLink>; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama.<br /><searchLink fieldCode="AU" term="%22Lindeman+B%22">Lindeman B</searchLink>; Department of Surgery, University of Alabama at Birmingham, Birmingham, Alabama. Electronic address: blindeman@uabmc.edu.
– Name: TitleSource
  Label: Source
  Group: Src
  Data: <searchLink fieldCode="JN" term="%220376340%22">The Journal of surgical research</searchLink> [J Surg Res] 2026 Oct; Vol. 326, pp. 525-543. <i>Date of Electronic Publication: </i>2026 Aug 05.
– Name: TypePub
  Label: Publication Type
  Group: TypPub
  Data: Journal Article; Evaluation Study
– Name: Language
  Label: Language
  Group: Lang
  Data: English
– Name: TitleSource
  Label: Journal Info
  Group: Src
  Data: <i>Publisher: </i><searchLink fieldCode="PB" term="%22Academic+Press%22">Academic Press </searchLink><i>Country of Publication: </i>United States <i>NLM ID: </i>0376340 <i>Publication Model: </i>Print-Electronic <i>Cited Medium: </i>Internet <i>ISSN: </i>1095-8673 (Electronic) <i>Linking ISSN: </i><searchLink fieldCode="IS" term="%2200224804%22">00224804 </searchLink><i>NLM ISO Abbreviation: </i>J Surg Res <i>Subsets: </i>MEDLINE
– Name: PublisherInfo
  Label: Imprint Name(s)
  Group: PubInfo
  Data: <i>Publication</i>: New York, NY : Academic Press<br /><i>Original Publication</i>: Philadelphia [etc.]
– Name: SubjectMESH
  Label: MeSH Terms
  Group: Su
  Data: <searchLink fieldCode="MM" term="%22Abstracting+and+Indexing%22">Abstracting and Indexing*</searchLink>/<searchLink fieldCode="MM" term="%22Abstracting+and+Indexing+methods%22">methods</searchLink> <br /><searchLink fieldCode="MM" term="%22Artificial+Intelligence%22">Artificial Intelligence*</searchLink><br /><searchLink fieldCode="MH" term="%22Pilot+Projects%22">Pilot Projects</searchLink> ; <searchLink fieldCode="MH" term="%22Feasibility+Studies%22">Feasibility Studies</searchLink> ; <searchLink fieldCode="MH" term="%22Generative+Artificial+Intelligence%22">Generative Artificial Intelligence</searchLink> ; <searchLink fieldCode="MH" term="%22Humans%22">Humans</searchLink> ; <searchLink fieldCode="MH" term="%22Large+Language+Models%22">Large Language Models</searchLink>
– Name: Abstract
  Label: Abstract
  Group: Ab
  Data: Introduction: Visual abstracts (VAs) have emerged as an efficient tool to disseminate scientific research, offering faster processing and improved information retention compared to traditional text-only abstracts. Concurrently, artificial intelligence (AI) has seen rapid advancements in language models and image generation. This study aims to evaluate the feasibility of using GPT-4o, and its more advanced model GPT-5.3, available on GPT-Pro paid version to generate VAs for scientific literature.<br />Methods: A pilot study was conducted using a dataset comprised of 26 pairs of visual and textual abstracts. The dataset was split into a training set (n = 20, 77%) and a testing set (n = 6, 23%) to approximate the commonly used 80-20 rule in machine learning while accommodating our sample size. GPT4o and GPT 5.3 were used as the AI models. The system was fine-tuned on the training set, with each input including the textual abstract and corresponding VA. In the testing phase, GPT4o and GPT 5.3 were provided the six-test set textual abstracts and a template generated from expert guidelines, prompting it to generate VAs. GPT 5.3 output was further refined with two iterative prompts after its first generated image. Our research team then compared the AI-generated VAs to the journal-created VAs, focusing on adherence to the template, accuracy of information representation, visual clarity, and overall design consistency.<br />Results: VAs produced by GPT4o demonstrated significant limitations. They consistently deviated from the provided template, exhibiting excessive imagery and text that often obscured the main message. All six generated VAs contained on average 17 instances (range 12-22) of misspelled and distorted text, impeding readability and comprehension. Notably, there was a lack of consistent structure across the AI-generated VAs, with each output varying considerably in format and style. Information accuracy was also compromised, with the AI output misrepresenting or omitting key data points from the original abstracts for the entire test cohort. The visual clarity was suboptimal, with cluttered layouts and poor color schemes that failed to effectively highlight the most important information. In contrast, VAs generated by GPT 5.3, before and after the iterative prompts, improved widely upon all four domains. While it still made errors regarding journal logo and author name, the rest of the design and consistency was comparable to the human-generated VAs.<br />Conclusions: Our pilot study demonstrates that while the GPT4o model lacks the capability to produce VAs suitable for scientific knowledge dissemination, a stark improvement was noticed with the most recent GPT-5.3 model, with its ability to adhere to the standardized template and produce images with clarity, accuracy, and visual consistency, required for effective research communication, rendering it useful for generation of a first draft of VAs.<br /> (Copyright © 2026 Elsevier Inc. All rights reserved.)
– Name: SubjectMinor
  Label: Contributed Indexing
  Group:
  Data: <i>Keywords: </i>Artificial intelligence; Image generation; Visual abstract
– Name: DateEntry
  Label: Entry Date(s)
  Group: Date
  Data: <i>Date Created: </i>20260805 <i>Date Completed: </i>20261008 <i>Latest Revision: </i>20261008
– Name: DateUpdate
  Label: Update Code
  Group: Date
  Data: 20261008
– Name: DOI
  Label: DOI
  Group: ID
  Data: 10.1016/j.jss.2026.05.087
– Name: AN
  Label: PMID
  Group: ID
  Data: 42556195
PLink https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=cmedm&AN=42556195
RecordInfo BibRecord:
  BibEntity:
    Identifiers:
      – Type: doi
        Value: 10.1016/j.jss.2026.05.087
    Languages:
      – Code: eng
        Text: English
    PhysicalDescription:
      Pagination:
        StartPage: 525
    Subjects:
      – SubjectFull: Pilot Projects
        Type: general
      – SubjectFull: Feasibility Studies
        Type: general
      – SubjectFull: Generative Artificial Intelligence
        Type: general
      – SubjectFull: Humans
        Type: general
      – SubjectFull: Large Language Models
        Type: general
      – SubjectFull: Abstracting and Indexing methods
        Type: general
      – SubjectFull: Artificial Intelligence
        Type: general
    Titles:
      – TitleFull: Evaluating the Feasibility of Artificial Intelligence in Generating Visual Abstracts: A Pilot Study.
        Type: main
  BibRelationships:
    HasContributorRelationships:
      – PersonEntity:
          Name:
            NameFull: Akhund R
      – PersonEntity:
          Name:
            NameFull: Mainali B
      – PersonEntity:
          Name:
            NameFull: Izhar A
      – PersonEntity:
          Name:
            NameFull: Muñoz Wilson DD
      – PersonEntity:
          Name:
            NameFull: Gillis A
      – PersonEntity:
          Name:
            NameFull: Fazendin JM
      – PersonEntity:
          Name:
            NameFull: Chen H
      – PersonEntity:
          Name:
            NameFull: Lindeman B
    IsPartOfRelationships:
      – BibEntity:
          Dates:
            – D: 01
              M: 10
              Text: 2026 Oct
              Type: published
              Y: 2026
          Identifiers:
            – Type: issn-electronic
              Value: 1095-8673
          Numbering:
            – Type: volume
              Value: 326
          Titles:
            – TitleFull: The Journal of surgical research
              Type: main
ResultId 1