Quality and Reliability of AI Information on Dental Implant Failure: A Comparative Multi-Model Analysis

Area of Science:

• Artificial Intelligence in Healthcare
• Dental Implantology
• Patient Education

Background:

• Dental implant failure is a significant concern for patients.
• AI chatbots offer a potential avenue for patient information dissemination.
• The performance of different AI models in providing accurate and clear medical information is largely uncharacterized.

Purpose of the Study:

• To develop a consensus-based set of patient questions regarding dental implant failure.
• To evaluate and compare the accuracy, clarity, reliability, and readability of responses from four leading AI chatbots: ChatGPT-4, DeepSeek-R1, Microsoft Copilot, and Google Gemini.

Main Methods:

• Twenty-three expert-validated questions were generated from consensus reports (EAO 2021, ICOI Pisa).
• Questions were submitted to each AI model under standardized conditions.
• Responses were assessed using CLEAR criteria, mGQS, accuracy scales, DISCERN items, and Flesch readability indices, with nonparametric tests for comparisons.

Main Results:

• Significant performance variability was observed among AI models.
• Google Gemini demonstrated the highest accuracy (P<0.001), while ChatGPT-4 showed the greatest reliability (DISCERN scores).
• Microsoft Copilot produced the most fluent responses, and DeepSeek-R1 offered the best readability, despite variations in linguistic complexity.

Conclusions:

• AI chatbots show promise as supplementary tools for patient education on dental implant failure.
• Each AI model possesses distinct strengths: Gemini for accuracy, ChatGPT-4 for reliability, Copilot for fluency, and DeepSeek-R1 for readability.
• Tailored guidance and ongoing development are crucial to optimize AI-generated patient information for clinical use and accessibility.