Related Experiment Video
Updated: Jan 9, 2026
13:44
Project-Based Learning Guidelines for Health Sciences Students: An Analysis with Data Mining and Qualitative Techniques
Published on: December 9, 2022
4.1K
Exploring artificial intelligence chatbots in pediatric fluoride education: a cross-sectional study
Nevra Karamüftüoğlu1, Ezgi Aydın Varol2, Cenkhan Bal3
1Department of Pediatric Dentistry, Gülhane Faculty of Dentistry, University of Health Sciences, Ankara, 06010, Turkey. nvrserbest@hotmail.com.
Scientific Reports
|November 29, 2025
Summary
ChatGPT-4.o demonstrated superior reliability and informational quality in providing fluoride information to parents compared to Gemini Pro and DeepSeek V3. This AI tool shows potential for caregiver education in pediatric dentistry, enhancing comprehension of fluoride use.
Area of Science:
- Artificial Intelligence in Healthcare
- Pediatric Dentistry
- Health Communication
Background:
- Large language model (LLM) chatbots are increasingly used in healthcare for accessible information.
- Fluoride is crucial for pediatric caries prevention but faces public concern and misinformation.
- Reliable digital communication is needed to address caregiver questions on fluoride use.
Purpose of the Study:
- To evaluate the performance of advanced AI chatbots (ChatGPT-4.o, Gemini Pro, DeepSeek V3) in delivering fluoride-related information for pediatric oral health.
- To assess the quality, reliability, readability, and originality of chatbot responses to caregiver queries.
Main Methods:
- An observational study presented 20 fluoride-related questions to three AI chatbots.
- Responses were evaluated by three blinded reviewers using validated tools (EQIP, DISCERN, GQS, FRES, FKRGL, iThenticate).
- Statistical analyses included ANOVA/Kruskal-Wallis tests and ICC for inter-rater reliability.
Main Results:
- ChatGPT-4.o significantly outperformed Gemini Pro and DeepSeek V3 in EQIP and DISCERN scores (p < 0.001), indicating superior reliability and informational quality.
- ChatGPT-4.o produced more readable and original content, though Flesch Reading Ease Score (FRES) and similarity index showed no significant differences.
- Flesch-Kincaid Reading Grade Level (FKRGL) differences were borderline and not statistically significant after correction; Global Quality Scale (GQS) outcomes were comparable.
Conclusions:
- ChatGPT-4.o demonstrated the clearest and most reliable communication regarding fluoride for pediatric oral health among the evaluated models.
- Its higher scores suggest potential as a supportive tool for caregiver education, enhancing comprehension of fluoride use.
- Cautious implementation with professional oversight and continuous validation is recommended to prevent misinformation and ensure safe clinical integration.
