Application of generative artificial intelligence models in diagnosis of localized skin diseases in a tertiary care teaching hospital in Pakistan
DOI:
https://doi.org/10.66344/jpad.v36i3.3387Keywords:
artificial intelligence, AI, Skin Diagnosis, ChatGPT, Gemini,GrokAbstract
Background Generative artificial neural networks like ChatGPT, Gemini, and Grok are rapidly being investigated for applications in medicine. Their application in dermatology, particularly for visual diagnostics, is still in its early phases.
Objective To assess the clinical usefulness and diagnostic precision of generative AI tools in identifying localized skin diseases from photographs, as well as the satisfaction of consultant dermatologists and suggestions for their application in clinical training.
Methods A total of 216 patients were enrolled with clinical images examined utilizing ChatGPT, Gemini,and Grok (72 patients each). Each AI tool made a diagnosis based only on the image input. Consultant dermatologists independently reviewed all patients and served as the gold standard. Consultant satisfaction ratings and recommendations for future use were also collected. SPSS was used to analyze the data, with Chi-square tests used to determine statistical significance (P<.05).
Results With a diagnosis accuracy of 61.1%, ChatGPT outperformed Gemini, (1.4%) and Grok (0%). Grok had the most missing or improper responses, whereas ChatGPT had the highest consultant satisfaction rate (54.2%). Chatgpt was recommended (79.1%) by consultant dermatologist in comparison with other AI tools. Variations were statistically significant (P<.01) in all domains: recommendation, accuracy, and satisfaction.
Conclusion In terms of clinical recommendation, consultant satisfaction, and diagnostic accuracy, ChatGPT performed better than Gemini, and Grok. Although more research involving clinical data is necessary, its image-based diagnostic capability shows potential for future usage in dermatology teaching and decision assistance.
References
1. Vallverdú J. Challenges and Controversies of Generative AI in Medical Diagnosis. Euphyía, 2023;17(32):88-121.
https://doi.org/10.33064/32euph4957
2. Akyon SH, Akyon FC, Camyar AS, Hızlı F, Sari T, Hızlı Ş. Evaluating the capabilities of generative AI tools in understanding medical papers: qualitative study. JMIR Med Inform. 2024;12:e59258. doi:10.2196/59258.
3. Shokrollahi Y, Yarmohammadtoosky S, Nikahd MM, Dong P, Li X, Gu L. A comprehensive review of generative AI in healthcare. arXiv [Preprint]. 2023 Oct 1 [cited 2026Sep24];arXiv:2310.00795. doi:10.48550/arXiv.2310.00795
4. Sengupta D. Artificial intelligence in diagnostic dermatology: challenges and the way forward. Indian Dermatol Online J. 2023;14(6):782-7. doi:10.4103/idoj.idoj_462_23. PMID: 38099026; PMCID: PMC10718130.
5. Gui H, Omiye JA, Chang CT, Daneshjou R. The promises and perils of foundation models in dermatology. J Invest Dermatol. 2024;144(7):1440-8. doi:10.1016/j.jid.2023.12.019. PMID:38441507.
6. Mondal H, Mondal S, Podder I. Using ChatGPT for writing articles for patients’ education for dermatological diseases: a pilot study. Indian Dermatol Online J. 2023;14(4):482-6. doi:10.4103/idoj.idoj_72_23. PMID:37521213; PMCID:PMC10373821.
7. Saeed M, Naseer A, Masood H, Rehman SU, Gruhn V. The power of generative AI to augment for enhanced skin cancer classification: a deep learning approach. IEEE Access.2023;11:130330-130344. doi:10.1109/ACCESS.2023.3332628
8. Sankar A, Chaturvedi K, Nayan AA, Hesamian MH, Braytee A, Prasad M. RETRACTED: Utilizing generative adversarial networks for acne dataset generation in dermatology. Bio Med Informatics. 2024;4(2):1059-70.
doi:10.3390/biomedinformatics4020059.
9. McCrary MR, Galambus J, Chen WS. Evaluating the diagnostic performance of a large language model-powered chatbot for providing immunohistochemistry recommendations in dermatopathology. J Cutan Pathol. 2024;51(9):689-695. doi:10.1111/cup.14631. Epub 2024 May 14. PMID:38744501.
10. Pillai J, Li B. Generative artificial intelligence in dermatology: Recommendations for future studies evaluating the clinical knowledge of models. Vol. 30, Skin Research and Technology. John Wiley and Sons Inc; 2024.
11. Singh JP. The impacts and challenges of generative artificial intelligence in medical education, clinical diagnostics, administrative efficiency, and data generation. Int J Appl Health Care Anal. 2023;8(5):37-46.
12. Manoharan P, Surapaneni KM. Assessing the diagnostic capability of ChatGPT through clinical case scenarios in dermatology. Indian J Dermatol Venereol Leprol. 2025;91(Suppl 1):S27-S28. doi:10.25259/IJDVL_1267_2023.
13. Habib MM, Hoodbhoy Z, Siddiqui MR. Knowledge, Attitudes, and Perceptions of Healthcare Students and Professionals on the Use of Artificial Intelligence in Healthcare in Pakistan. PLOS Digital Health. 2024 May;3(5):e0000443.
doi: 10.1371/journal.pdig.0000443.
14. Hirosawa T, Harada Y, Yokose M, Sakamoto T, Kawamura R, Shimizu T. Diagnostic accuracy of differential-diagnosis lists generated by generative pretrained transformer 3 chatbot for clinical vignettes with common chief complaints: a pilot study. Int J Environ Res Public Health. 2023;20(4):3378.
15. Johnson D, Goodman R, Patrinely JR, Stone CA, Zimmerman E, Donald R, et al. Assessing the accuracy and reliability of AI-generated medical responses: an evaluation of the Chat-GPT model. Res Sq [Preprint]. 2023 Feb 28:rs.3.rs-2566942. doi:10.21203/rs.3.rs-2566942/v1.
16. Kuroiwa T, Sarcon A, Ibara T, Yamada E, Yamamoto A, Tsukamoto K, Fujita K. The potential of ChatGPT as a self-diagnostic tool in common orthopedic diseases: exploratory study. J Med Internet Res. 2023;25:e47621. doi:10.2196/47621
Published
Issue
Section
License
Copyright (c) 2026 Zafarullah Khan, Samrat Habib, Rabeeka bakhtiar, Khawar Khurshid, Aisha Ghias, Ali Nawaz, Taha Habib
This work is licensed under a Creative Commons Attribution 4.0 International License.
Submission declaration
Authors retain the copyright to their work and grant the 'Journal of Pakistan Association of Dermatologists (JPAD)' the right of first publication under a Creative Commons Attribution 4.0 International (CC BY 4.0) license. This license allows others to share, adapt, and reuse the work for any purpose, including commercial use, as long as appropriate credit is given to the original authors and the journal.
By submitting a manuscript, authors confirm that the work has not been published previously (except as an abstract, lecture, or academic thesis), is not under review elsewhere, and has been approved by all authors and relevant authorities. Once accepted, the article will be openly accessible under the CC BY 4.0 license, ensuring wide dissemination and reuse with proper attribution.
