YAPAY ZEKÂ MODELLERİ DİŞ BEYAZLATMADA KLİNİK KARAR VERME SÜREÇLERİNE HAZIR MI? CHATGPT-5, GEMİNİ 2.5 PRO, DEEPSEEK V3.2 VE CLAUDE SONNET 4.5’İN DOĞRULUK, TAMLIK VE OKUNABİLİRLİK AÇISINDAN KANITA DAYALI DEĞERLENDİRİLMESİ

Öz

Amaç: Bu çalışmanın amacı, büyük dil modelleri (LLM) olan ChatGPT-5, Gemini 2.5 Pro, DeepSeek v3.2 ve Claude Sonnet 4.5 tarafından diş hekimliğinde beyazlatma ile ilgili sunulan bilgilerin doğruluk, tamlık ve okunabilirlik açısından karşılaştırılmasıdır. Yöntem: Diş beyazlatmasının çeşitli yönlerini kapsayan toplam 25 açık uçlu soru hazırlanmış ve dört farklı büyük dil modeline yöneltilmiştir. Elde edilen yanıtlar kaydedilmiş ve kaynakları gizlenmiş şekilde iki restoratif diş tedavisi uzmanı tarafından bağımsız olarak değerlendirilmiştir. Doğruluk ve kapsamlılık sırasıyla 5’li ve 3’lü Likert ölçekleri kullanılarak değerlendirilmiştir. Okunabilirlik ise Flesch Reading Ease Score (FRES), Flesch–Kincaid Grade Level (FKGL) ve Simple Measure of Gobbledygook (SMOG) indeksleri kullanılarak analiz edilmiştir. İstatistiksel analizlerde Sınıf İçi Korelasyon Katsayısı (ICC), Shapiro–Wilk testi, Kruskal–Wallis testi, Dunn–Bonferroni testi ve Spearman korelasyon analizi kullanılmıştır. Bulgular: Diş beyazlatmaya ilişkin açık uçlu sorulara verilen yanıtların doğruluk puanları açısından yapay zekâ modelleri arasında istatistiksel olarak anlamlı farklılık saptanmıştır (p<0,05). DeepSeek v3.2’nin doğruluk puanı (4,52±0,77), ChatGPT-5’in doğruluk puanından (3,88±1,01) istatistiksel olarak anlamlı derecede daha yüksek bulunmuştur. Okunabilirlik açısından değerlendirilen FRES puanlarının ise DeepSeek v3.2’de (42,02±7,63), ChatGPT-5 (29,09±9,07), Gemini 2.5 Pro (33,92±9,24) ve Claude Sonnet 4.5’e (21,65±9,29) kıyasla istatistiksel olarak anlamlı derecede daha yüksek olduğu belirlenmiştir. Sonuç: DeepSeek v3.2, doğruluk açısından ChatGPT-5’e kıyasla daha iyi performans göstermiştir. Yanıtların tamlık puanları bakımından modeller arasında istatistiksel olarak anlamlı bir farklılık bulunmamıştır. Bununla birlikte, FRES sonuçları DeepSeek v3.2’nin diğer sohbet botlarına göre daha yüksek okunabilirliğe sahip metinler ürettiğini göstermiştir. Genel olarak, bu bulgular DeepSeek v3.2’nin hem doğruluk hem de okunabilirlik açısından üstün performans sergilediğini ortaya koymaktadır.

Anahtar Kelimeler

• Yapay zekâ
• büyük dil modelleri
• diş beyazlatma
• diş hekimliği

Etik Beyan

Bu çalışma, Helsinki Bildirgesi'nin etik ilkeleri doğrultusunda yürütülmüş olup Etik Kurul onayı gerektirmemektedir.

ARE AI MODELS READY FOR CLINICAL DECISION-MAKING IN DENTAL BLEACHING? AN EVIDENCE-BASED EVALUATION OF ACCURACY, COMPLETENESS, AND READABILITY OF CHATGPT-5, GEMINI 2.5 PRO, DEEPSEEK V3.2, AND CLAUDE SONNET 4.5

Öz

Objective: The aim of this study was to compare the accuracy, completeness, and readability of information related to whitening in dentistry provided by the large language models (LLM) ChatGPT-5, Gemini 2.5 Pro, DeepSeek v3.2, and Claude Sonnet 4.5. Methods: A total of 25 open-ended questions covering various aspects of dental whitening were prepared and presented to four different LLMs. The responses were recorded and evaluated independently by two restorative dental treatment specialists who were blinded to the source of the responses. Accuracy and completeness were evaluated using 5-point and 3-point Likert scales respectively. Readability was evaluated using the Flesch Reading Ease Score (FRES), the Flesch–Kincaid Grade Level (FKGL), and the Simple Measure of Gobbledygook (SMOG) index. In the statistical analyses, the Intraclass Correlation Coefficient (ICC), Shapiro-Wilk test, Kruskal-Wallis test, Dunn-Bonferroni test, and Spearman correlation analysis were used. Results: A statistically significant difference was identified among the artificial intelligence models regarding the accuracy of their responses to open-ended questions related to dental whitening (p<0.05). The accuracy score of DeepSeek v.3.2 (4.52±0.77) was determined to be statistically significntly higher than that of ChatGPT-5 (3.88±1.01). The FRES points for readability were found to be statistically significantly higher for DeepSeek v3.2 (42.02±7.63) compared to the mean points obtained for ChatGPT-5 (29.09±9.07), Gemini 2.5 Pro (33.92±9.24), and Claude Sonnet 4.5 (21.65±9.29). Conclusion: DeepSeek v3.2 exhibited better performance than ChatGPT-5 in terms of accuracy. No statistically significant differences were identified among the models regarding the completeness scores of responses to open-ended questions. However, the FRES results indicated that DeepSeek v3.2 generated texts with higher readability compared to the other chatbots. Overall, these findings suggest that DeepSeek v3.2 demonstrated superior performance in terms of both accuracy and readability.

Anahtar Kelimeler

• Artificial intelligence
• large language models
• teeth whitening
• dentistry

Etik Beyan

This study was conducted in accordance with the ethical principles of the Helsinki Declaration and did not require Ethics Committee approval.

Kaynakça

1. Thurzo A, Urbanova W, Novak B, Czako L, Siebert T, Stano P, et al. Where is artificial intelligence applied in dentistry? A systematic review and literature analysis. Healthcare. 2022;10:1234. doi:10.3390/healthcare10071269
2. Acar AH. Can natural language processing serve as a consultant in oral surgery? J Stomatol Oral Maxillofac Surg. 2023;125:101724. doi:10.1016/j.jormas.2023.101724
3. Dursun D, Bilici Geçer R. Can artificial intelligence models serve as patient information consultants in orthodontics? BMC Med Inform Decis Mak. 2024;24:211. doi:10.1186/s12911-024-02619-8
4. Yu E, Chu X, Zhang W, et al. Large language models in medicine: applications, challenges, and future directions. Int J Med Sci. 2025;22:2792–2801. doi:10.7150/ijms.111780
5. Xu X, Chen Y, Miao J. Opportunities, challenges, and future directions of large language models, including ChatGPT, in medical education: a systematic scoping review. J Educ Eval Health Prof. 2024;21:6. doi:10.3352/jeehp.2024.21.6
6. Eggmann F, Blatz MB. ChatGPT: Chances and challenges for dentistry. Compend Contin Educ Dent. 2023;44(4):220–224.
7. Taşyürek M, Adıgüzel Ö, Gündoğar M, Goncharuk-Khomyn M, Ortaç H. Comparative evaluation of the responses from ChatGPT-5, Gemini 2.5 Flash, and DeepSeek-V3.1 chatbots to patient inquiries about endodontic treatment in terms of accuracy, understandability, and readability. International Dental Research. 2025;15(3):91–95. doi:10.5577/intdentres.662.
8. İlter Er Ö, Adıgüzel Ö, Taşyürek M, Ortaç H. Comparison of ChatGPT-5, Gemini 2.5 Pro, and DeepSeek-V3.1 chatbot responses to dental avulsion injuries. Health Sci Monit. 2025;3:e251201. doi:10.5577/hesmon.e251201

9. Carlà MM, et al. Testing the power of Google DeepMind: Gemini versus ChatGPT-4 facing a European ophthalmology examination. AJO Int. 2024;1:100063. doi:10.1016/j.ajoint.2024.100063
10. Masalkhi M, Ong J, Waisberg E, Lee AG. Google DeepMind’s Gemini AI versus ChatGPT: a comparative analysis in ophthalmology. Eye. 2024;38:1412–1417. doi:10.1038/s41433-024-02958-w
11. Imran M, Almusharraf N. Google Gemini as a next-generation AI educational tool: a review of emerging educational technology. Smart Learn Environ. 2024;11:22. doi:10.1186/s40561-024-00310-z
12. Team G, Anil R, Borgeaud S, Wu Y, Alayrac JB, Yu J, et al. Gemini: a family of highly capable multimodal models. 2023. doi:10.48550/arXiv.2312.11805
13. DeepSeek AI. DeepSeek-V3 technical report. arXiv. 2025; arXiv:2412.19437.
14. Ye J, et al. DeepSeek in healthcare: a survey of capabilities, risks, and clinical applications of open-source large language models. arXiv. 2025; arXiv:2506.01257. doi:10.48550/arXiv.2506.01257
15. Deng Z, Han Q, Zhou W, Zhu X, Wen S, Xiang Y. Exploring DeepSeek: advances, applications, challenges, and future directions. IEEE/CAA J Autom Sin. 2025;12:872–893. doi:10.1109/JAS.2025.125498
16. Jin H, et al. Comparative study of Claude 3.5 Sonnet and human physicians in generating discharge summaries. Front Digit Health. 2024;6:1456911.doi: 10.3389/fdgth.2024.1456911
17. Nasirov R. The role of Claude 3.5 Sonnet and ChatGPT-4 in posterior cervical fusion patient guidance. World Neurosurg. 2025;197:123889. doi:10.1016/j.wneu.2025.123889
18. Anthropic. Introducing Claude Sonnet 4.5 [Internet]. 2025 [cited 2025 Nov 10]. Available from: https://www.anthropic.com
19. Anthropic. Claude Sonnet: Model overview and performance improvements [Internet]. 2025 [cited 2025 Nov 10]. Available from: https://www.anthropic.com
20. Carey CM. Tooth whitening: what we now know. J Evid Based Dent Pract. 2014;14:70–76. doi:10.1016/j.jebdp.2014.02.006
21. Epple M, Meyer F, Enax J. A critical review of modern concepts for teeth whitening. Dent J. 2019;7:79. doi:10.3390/dj7030079
22. Silva EMD, et al. Can whitening toothpastes maintain optical stability of enamel over time? J Appl Oral Sci. 2018;26:e20160460. doi:10.1590/1678-7757-2016-0460
23. Ubaldini AL, et al. Hydrogen peroxide diffusion dynamics in dental tissues. J Dent Res. 2013;92:661–665. doi:10.1177/0022034513488893
24. Pauli MC, et al. Current status of whitening agents and enzymes in dentistry. Braz J Pharm Sci. 2022;58:e19501. doi:10.1590/s2175979020201000X32e19501
25. Kihn PW. Vital tooth whitening. Dent Clin North Am. 2007;51:319–331. doi:10.1016/j.cden.2006.12.001
26. Simonsen RJ. Ultraconservative treatment options. In: Biomimetic restorative dentistry. Berlin: Quintessence; 2021. p. 45–67.
27. Hillard JJ, Robinson PB. Additional conservative esthetic procedures. In: Sturdevant’s art and science of operative dentistry. 7th ed. St. Louis (MO): Elsevier; 2019. p. 235–260.
28. Haywood VB. Bleaching procedures. In: Cohen’s pathways of the pulp. 12th ed. St. Louis (MO): Elsevier; 2020. p. 2970–3038.
29. Likert R. A technique for the measurement of attitudes. Arch Psychol. 1932;22(140):1–55.
30. Flesch R. A new readability yardstick. J Appl Psychol. 1948;32(3):221–233. doi:10.1037/h0057532
31. Esmailpour H, et al. Performance of artificial intelligence chatbots in responding to patient FAQs on dental prostheses. BMC Oral Health. 2025;25:574. doi:10.1186/s12903-025-05965-9
32. Daraz L, et al. Readability of online health information: a meta-narrative systematic review. Am J Med Qual. 2018;33:487–492. doi:10.1177/1062860617751639
33. Bulut AC, Bahadır HS, Ateş G. Artificial intelligence in dental education: can AI-based chatbots compete with general practitioners? BMC Med Educ. 2025;25:1319. doi:10.1186/s12909-025-07880-7
34. Özcivelek T, Özcan B. Comparative evaluation of responses from DeepSeek-R1, ChatGPT-o1, ChatGPT-4, and Dental GPT chatbots. BMC Oral Health. 2025;25:871. doi:10.1186/s12903-025-06267-w
35. Diniz-Freitas M, Diz-Dios P. DeepSeek: another step forward in the diagnosis of oral lesions. J Dent Sci. 2025;20:1904–1907. doi:10.1016/j.jds.2025.02.023
36. Performance of ChatGPT-4, Gemini, and DeepSeek-V3 on dental licensing examination questions. J Dent Sci. 2025;20:2154–2162. doi:10.1016/j.jds.2025.07.011
37. Zhang H, et al. DeepSeek performs better than other large language models in dental cases. arXiv. 2025;arXiv:2509.02036. doi:10.48550/arXiv.2509.02036
38. Suárez A, et al. Assessing ChatGPT as an intelligent virtual assistant in oral surgery. Comput Struct Biotechnol J. 2023;24:46–52. doi:10.1016/j.csbj.2023.11.058
39. Gurbuz E, Tetik B. Accuracy and completeness of ChatGPT-4o in the management of non-carious cervical lesions. Digit Dent J. 2025;2:100015. doi:10.1016/j.ddj.2025.100015
40. American Academy of Pediatric Dentistry. Policy on the use of dental bleaching for child and adolescent patients. In: The Reference Manual of Pediatric Dentistry. Chicago (IL): American Academy of Pediatric Dentistry; 2025. p. 136-140. https://www.aapd.org/globalassets/media/policies_guidelines/p_bleaching25.pdf?utm_source=chatgpt.com
41. Jacobs T, et al. Is ChatGPT an accurate and readable patient aid for third molar extractions? J Oral Maxillofac Surg. 2024;82:1239–1245. doi:10.1016/j.joms.2024.06.177
42. Balel Y. Can ChatGPT be used in oral and maxillofacial surgery? J Stomatol Oral Maxillofac Surg. 2023;124:101471. doi:10.1016/j.jormas.2023.101471
43. Friedman DB, Hoffman-Goetz L. A systematic review of readability and comprehension instruments used for health information. Health Educ Behav. 2006;33:352–373. doi:10.1177/1090198105277329
44. Das D, et al. AI chatbots in ocular oncology: a comparative study. Cureus. 2025;17:e90773. doi:10.7759/cureus. 90773
45. Taşyürek M, Adıgüzel Ö, Ortaç H. Comparative evaluation of ChatGPT-5, Gemini 2.5 Flash, Grok 4, and Claude Sonnet-4 in endodontic iatrogenic events. Healthcare. 2025;13:2615. doi:10.3390/healthcare13202615
46. Meng X, et al. The application of large language models in medicine: a scoping review. iScience. 2024;27:109713. doi:10.1016/j.isci.2024.109713
47. Hoyt RE, et al. Evaluating large reasoning model performance on complex medical scenarios. medRxiv. 2025;2025.04. doi:10.1101/2025.04.29.25326666
48. American Medical Association. AMA issues new principles for AI development, deployment & use [Internet]. Chicago (IL): American Medical Association; 2023 Nov 28 [cited 2026 Jun 28]. https://www.ama-assn.org/press-center/ama-press-releases/ama-issues-new-principles-ai-development-deployment-useutm_source=chatgpt.com
49. Moreno AC, Bitterman DS. Toward clinical-grade evaluation of large language models. Int J Radiat Oncol Biol Phys. 2024;118:916–920. doi:10.1016/j.ijrobp.2023.11.012
50. Tang L, et al. Evaluating large language models on medical evidence summarization. medRxiv. 2023. doi:10.1101/2023.04.22.23288967
51. Johri S, et al. Guidelines for rigorous evaluation of clinical LLMs for conversational reasoning. medRxiv. 2023. doi:10.1101/2023.09.12.23295399
52. Aguiar de Sousa R, et al. Is ChatGPT a reliable source of scientific information regarding third-molar surgery? J Am Dent Assoc. 2024;155:227–232.e6. doi:10.1016/j.adaj.2023.11.004
53. Pan A, Musheyev D, Bockelman D, Loeb S, Kabarriti AE. Assessment of artificial intelligence chatbot responses to top searched queries about cancer. JAMA Oncol. 2023;9:1437–1440. doi:10.1001/jamaoncol.2023.2947