ArticleScientific reports2026
Performance, safety, and limitations of multimodal large language models in wound image assessment.
Article in Scientific reports, 2026. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Not yet cited in PubMed.
What it found
Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.
The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.
The trial behind it
Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.
Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.
Who cites it
0 citing papers in PubMed.
No citing paper in PubMed yet.
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
6 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
Accurate visual assessment of acute, chronic, and surgical wounds is fundamental to clinical decision-making and is increasingly performed using digital wound photographs in routine and remote care. However, interpretation of wound images remains subjective and variable across clinicians, and the rapid emergence of vision-capable large language models (LLMs) has led to their informal use for wound description and clinical interpretation despite limited task-specific validation. To address this gap, we compared three advanced vision-capable LLMs (ChatGPT-4o, Claude 3.5 Sonnet, and Gemini Advanced) for structured wound image assessment using standardized clinical frameworks. From 1,200 clinical wound photographs obtained during routine care, 450 images (150 acute, 150 chronic, and 150 surgical) were randomly selected and independently reviewed by three expert clinicians to establish expert consensus reference standards. Each model received identical prompts incorporating the Bates-Jensen Wound Assessment Tool (BWAT) and the TIMES framework, and outputs were evaluated for diagnostic accuracy, assessment quality, treatment recommendation appropriateness, safety, understandability, and actionability; informational quality was assessed using DISCERN and PEMAT-P. ChatGPT-4o achieved the highest accuracy for diagnosis (51.3%), clinical conclusion (52.4%), and treatment recommendations (53.6%), while Claude performed best for wound sizing (72.0%) and urgency determination (70.9%). Gemini showed substantial limitations, with non-response in 67-68% across several assessment domains and the lowest clinical and safety performance. Overall, ChatGPT-4o most consistently generated accurate, structured, and clinically aligned wound assessments. However, the findings also highlight important limitations in reliability, safety, and real-world applicability, underscoring the need for further validation and explicit safeguards before routine clinical integration.
Indexed as
Identifiers
What OpenQuestion holds
Registered trials
Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.