EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Evaluation of ChatGPT-4 in detecting referable diabetic retinopathy using single fundus images

Owais Aftab, Hamza Khan, Brian L. VanderBeek, Drew Scoles, Benjamin J. Kim, Jonathan C. Tsui

AJO International · 2025-07 · 2 (2) · 100111

DOI: 10.1016/j.ajoint.2025.100111 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AJO International, 2025-07. Verfasst von Owais Aftab und 5 weiteren Autor:innen. Die Arbeit wurde bisher 10-mal zitiert (Crossref) und nennt 22 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

• ChatGPT-4 is unable to adequately read and interpret normal and mild NPDR images limiting utility in low-risk settings. • ChatGPT-4 shows high sensitivity for identifying referable diabetic retinopathy but lacks specificity. • ChatGPT-4 needs improvement in multimodal image fund of knowledge in order to improve real-world applicability. Evaluate ChatGPT-4′s ability to identify referable diabetic retinopathy (DR) from single fundus images. A cross-sectional study comparing ChatGPT-4′s versus retina specialists’ identification of more than mild DR (mtmDR) and vision-threatening DR (VTDR). Images in equal proportions of normal, mild, moderate, and severe nonproliferative DR (NPDR), proliferative DR (PDR), and blurry images with and without suspected PDR were presented to a panel of blinded retina specialists who identified images as readable or unreadable, and potentially as mtmDR or VTDR. These images were also submitted to ChatGPT-4 three times with a standardized prompt regarding mtmDR and VTDR. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for ChatGPT-4′s responses regarding mtmDR and VTDR as compared to the retina specialists majority determination. Retina specialists read 158/180 prompts (87.7 %) with excellent interrater reliability while ChatGPT-4 read 132/180 (73.33 %) of the image prompts. For mtmDR, ChatGPT-4 demonstrated a sensitivity of 96.2 %, specificity of 19.1 %, PPV of 69.1 %, and NPV of 72.7 %. Overall, 90.9 % of prompts read by ChatGPT-4 were labeled as mtmDR. For VTDR, ChatGPT-4 demonstrated a 63.0 % sensitivity, 62.5 % specificity, 71.9 % PPV, and 52.6 % NPV compared to retina specialists. ChatGPT-4 labeled 51.5 % of read images as VTDR. Overall referability was 66.6 % for retina specialists and 93.3 % for ChatGPT-4. While ChatGPT-4 demonstrates promise in identifying moderate-to-severe DR, its limited specificity and tendency to overcall disease reduce its current utility as a screening tool.

Zitieren

APA

Aftab, O., Khan, H., VanderBeek, B. L., Scoles, D., Kim, B. J., & Tsui, J. C. (2025). Evaluation of ChatGPT-4 in detecting referable diabetic retinopathy using single fundus images. AJO International, 2(2), 100111. https://doi.org/10.1016/j.ajoint.2025.100111

Vancouver

Aftab O, Khan H, VanderBeek BL, Scoles D, Kim BJ, Tsui JC. Evaluation of ChatGPT-4 in detecting referable diabetic retinopathy using single fundus images. AJO International. 2025;2(2):100111. doi:10.1016/j.ajoint.2025.100111

BibTeX
@article{Aftab2025,
  author = {Aftab, Owais and Khan, Hamza and VanderBeek, Brian L. and Scoles, Drew and Kim, Benjamin J. and Tsui, Jonathan C.},
  title = {Evaluation of ChatGPT-4 in detecting referable diabetic retinopathy using single fundus images},
  journal = {AJO International},
  year = {2025},
  volume = {2},
  number = {2},
  pages = {100111},
  doi = {10.1016/j.ajoint.2025.100111}
}
Kostenlos starten →