EUVIMEDEUVIMED entdecken
Open Access · CC BY-ND 4.0

Evaluation of Artificial Intelligence in Answering Dermatological Medical Questions

Abdullah ALSarrajie, Akram ALsubari

International Journal on Advanced Electrical and Computer Engineering · 19.01.2026 · 15 (1S) · 24-41

DOI: 10.65521/ijaece.v15i1s.1341 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal on Advanced Electrical and Computer Engineering, online erschienen am 19.01.2026. Verfasst von Abdullah ALSarrajie und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This research presents a systematic evaluation of the adaptation of a large Arabic linguistic model (based on AraGPT2) to answer questions in the field of dermatology. The study aims to bridge the gap in specialized linguistic resources by fine-tuning the model to a newly created and purified dataset, collected using a hybrid methodology combining web scraping and filtered data enrichment. This dataset consists of 40,132 specialized question-answer pairs. The performance of the finely tuned model was quantitatively assessed using BERTScore, BLEU, Levenshtein distance, and two types of initial human evaluation. The quantitative results showed strong semantic performance, with the model achieving a BERTScore (F1) of 64.49%, confirming its ability to effectively understand medical context and meaning. In contrast, the verbal matching measures reflected a tendency toward free generation, scoring BLEU at 10.00% and Levenshtein distance at 28.13%. These results demonstrate that the model favors the free generation of new formulations over the verbatim retrieval of reference texts. Furthermore, the qualitative results of the proposed model showed a competitive overall performance of 4.01, achieving a high score of 4.55 in the criteria of linguistic clarity and readability for non-expert audiences. These results confirm that the model's primary contribution lies in its ability to enhance human comprehension (understanding) of complex medical data. The study also underscores the urgent need for subsequent clinical validation by field experts to ensure complete clinical accuracy and reliability.

Zitieren

APA

ALSarrajie, A., & ALsubari, A. (2026). Evaluation of Artificial Intelligence in Answering Dermatological Medical Questions. International Journal on Advanced Electrical and Computer Engineering, 15(1S), 24-41. https://doi.org/10.65521/ijaece.v15i1s.1341

Vancouver

ALSarrajie A, ALsubari A. Evaluation of Artificial Intelligence in Answering Dermatological Medical Questions. International Journal on Advanced Electrical and Computer Engineering. 2026;15(1S):24-41. doi:10.65521/ijaece.v15i1s.1341

BibTeX
@article{ALSarrajie2026,
  author = {ALSarrajie, Abdullah and ALsubari, Akram},
  title = {Evaluation of Artificial Intelligence in Answering Dermatological Medical Questions},
  journal = {International Journal on Advanced Electrical and Computer Engineering},
  year = {2026},
  volume = {15},
  number = {1S},
  pages = {24--41},
  doi = {10.65521/ijaece.v15i1s.1341}
}
Kostenlos starten →