ArticleClinical ophthalmology (Auckland, N.Z.)2025
Diagnostic Accuracy and Counseling Quality of GPT-4o for Strabismus and Pseudostrabismus in Patient-Generated Mobile Photographs: A Preliminary Evaluation.
Article in Clinical ophthalmology (Auckland, N.Z.), 2025. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Cited by 1 paper.
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
1 citing paper in PubMed.
- Beyond GPT-4o: Interpreting AI's Role and Trust in Ophthalmic Care [Letter].Clinical ophthalmology (Auckland, N.Z.) · 2025Article
Corrections and comments
PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.
Authors and funding
4 authors.
Funding
No grant is acknowledged in the PubMed record.
Abstract
Background: Research-grade artificial intelligence has been used to accurately diagnose strabismus from image input. OpenAI's consumer-oriented GPT-4o model can analyze images, but has shown poor accuracy for image-based diagnosis. Parents may turn to GPT-4o to support or refute visible health concerns for their children, such as strabismus. The study aims to evaluate GPT-4o's diagnostic accuracy and response quality for strabismus evaluation. Methods: After gold-standard alternate cover exam by a clinician, 35 mobile photos of esotropia (13), pseudoesotropia (11), and exotropia (11) were selected. Images were excluded if a second "masked" examiner did not corroborate diagnosis. Images were submitted to a secure GPT-4o platform with patient-perspective prompts requesting overall evaluation (Prompt 1) and eye alignment evaluation (Prompt 2). Responses were graded by a pediatric ophthalmologist and certified orthoptist assessing for quality and safety. Results: GPT-4o provided interpretations for 15/35 and 27/35 images after Prompts 1 and 2, respectively. Analysis of the accuracy includes a primary "intention-to-diagnose" and secondary "per-diagnosis" framework. The diagnostic accuracies in the primary and secondary analysis following prompt 1 were 14.3% (low sensitivity and specificity) and 33.3% (low sensitivity, high specificity), respectively. Following prompt 2, accuracies were 48.6% (moderate sensitivity, low specificity) and 63.0% (high sensitivity, low specificity). Overall, the mean rating for content for strabismus prompts was 4.94 ± 0.27 out of a best possible 6, and for pseudostrabismus, 5.14 ± 0.18 (p=0.638). Conclusion: GPT-4o shows poor accuracy for image-based strabismus diagnosis. GPT-4o frequently categorized pseudoesotropia as true strabismus and true strabismus as orthophoria. While the quality of responses was rated as good overall, the quality of counseling did not match what would be provided in a pediatric ophthalmology clinic. Patients, clinicians, and AI developers should be aware of the need for specialist evaluation for strabismus.
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.