文献速递口腔正畸

患者对人工智能提供正畸建议的信任:系统综述

Patient trust in artificial intelligence for orthodontic advice: A systematic review.
PubMed 文献 · J Am Dent Assoc · 2025-09-26

该系统综述检索2022年1月1日至2025年2月28日五个数据库的横断面研究,评估AI聊天机器人回答正畸患者问题的可靠性。纳入13项研究,ChatGPT准确率分别为68.7%和92.6%,Bard和Gemini结果不一,AI不能替代专业咨询但可辅助回答常规问题。

要点

  • ChatGPT回答正畸问题的准确率分别为68.7%和92.6%。
  • Bard和Gemini表现不一,部分研究评分较低。
  • AI可辅助回答常规问题,但最终决策仍需专业医师。

核心结论

  • 大型语言模型尤其是ChatGPT可提供总体可靠的正畸信息,但不能替代专业咨询,最终诊疗决策仍应由合格专业人员做出。

研究设计与样本

  • 系统综述检索PubMed、Web of Science、Scopus、Ovid MEDLINE和Embase,纳入2022年1月1日至2025年2月28日发表的横断面研究,共纳入13项研究。

主要发现

  • ChatGPT准确率分别为68.7%和92.6%,在Likert量表和DISCERN评估中得分良好。
  • 部分研究报告较低评分(2.2/4.0、3.8/10.0),Bard和Gemini结果不一,显示聊天机器人性能存在不一致。

临床怎么用

  • AI聊天机器人可用于回答正畸常规患者问题,帮助管理常见咨询、减少临床工作量并提高成本效益,但最终诊疗决策仍由合格专业人员负责。

以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。

原文

BACKGROUND: Artificial intelligence (AI) large language models are being used increasingly for health care-related queries. This systematic review assessed the reliability of AI chatbot responses to patient questions related to orthodontics. TYPES OF STUDIES REVIEWED: PubMed, Web of Science, Scopus, Ovid MEDLINE, and Embase were systematically searched for studies published from January 1, 2022, through February 28, 2025. Only cross-sectional studies were included. Data were screened and extracted, and the risk of bias was assessed. RESULTS: Of 105 identified studies, 13 met the inclusion criteria. ChatGPT (OpenAI) showed 68.7% and 92.6% accuracy rates, scoring well on Likert scales and DISCERN evaluations. However, some studies reported lower scores (2.2/4.0, 3.8/10.0) and mixed results for Bard (Google) and Gemini (Google), highlighting inconsistencies in chatbot performance. PRACTICAL IMPLICATIONS: Whereas large language models, particularly ChatGPT, cannot replace professional consultations, they offer generally reliable orthodontic information and can be valuable tools for addressing routine patient questions. Based on evolving commentary, AI is likely to serve as a frontline support system in the future, helping manage common inquiries, reduce clinician workload, and improve cost-effectiveness while ensuring that final care decisions remain with qualified professionals. This review was registered in Open Science Factory (https://doi.org/10.17605/OSF.IO/X8HJT).

来源:J Am Dent Assoc,查看原文。版权归原作者所有。

引用

  1. Santhosh Kumar S, Michelogiannakis D, Xu X, et al. Patient trust in artificial intelligence for orthodontic advice: A systematic review. J Am Dent Assoc. 2025. PMID: 41014309. doi:10.1016/j.adaj.2025.08.003

供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。