沙特阿拉伯耳鼻咽喉科研究中的AI：当前格局、差距与未来方向的范围综述

文献信息

中文摘要

背景： 人工智能（AI）在耳鼻咽喉科中的应用日益增多，但沙特阿拉伯AI研究的范围和重点仍不明确。本范围综述梳理了现有证据，并识别了关键研究主题和知识空白。
方法： 按照PRISMA-ScR指南开展范围综述。检索PubMed/MEDLINE、Scopus、Web of Science、Google Scholar和沙特医学杂志档案，纳入2019年1月至2026年4月发表的研究。符合条件的研究为在沙特耳鼻咽喉科范围内探讨AI应用或临床医生观点的研究。证据按主要耳鼻咽喉科领域进行映射，包括人工耳蜗植入与听力、鼻科与颅底外科、面部整形外科、喉科与嗓音、头颈肿瘤学、阻塞性睡眠呼吸暂停的外科及其他耳鼻咽喉相关干预、AI教育与实施，以及AI辅助患者教育。
结果： 检索共识别2,925条数据库记录、4条其他来源记录以及7条来自更新检索和专家建议的记录；13项研究符合纳入标准。证据涵盖AI认知与教育、基于大语言模型（LLM）的患者教育、人工耳蜗植入、鼻科与颅底外科、面部整形外科、喉科与嗓音以及头颈肿瘤学；没有符合条件的研究涉及阻塞性睡眠呼吸暂停的外科或其他耳鼻咽喉相关干预。临床医生对AI总体持积极态度，而临床研究显示AI模型性能良好。LLM聊天机器人通常能给出准确的术后建议，但当使用经过验证的工具评估时，其回答较难阅读，且可操作性低于患者教育标准所建议的水平。然而，AI教育有限、数据集规模小或非沙特数据集、外部验证有限以及实施障碍仍是重要挑战。
结论： 沙特耳鼻咽喉科的AI研究仍然有限，但正在多个临床领域扩展。未来研究应优先关注沙特本土数据集、多中心临床验证、结构化AI教育、对LLM生成患者信息进行严格的双语（阿拉伯语和英语）评估，以及按照沙特2030愿景将AI安全整合到临床实践中。

英文摘要

BACKGROUND: Artificial intelligence (AI) is increasingly used in otolaryngology, but the extent and focus of AI research in Saudi Arabia remain unclear. This scoping review mapped the available evidence and identified key research themes and knowledge gaps.
METHODS: A scoping review was conducted following the PRISMA-ScR guidelines. PubMed/MEDLINE, Scopus, Web of Science, Google Scholar, and the Saudi Medical Journal archive were searched for studies published between January 2019 and April 2026. Eligible studies investigated AI applications or clinician perspectives within Saudi otolaryngology. The evidence was mapped across major otolaryngology domains, including cochlear implantation and hearing, rhinology and skull base surgery, facial plastic surgery, laryngology and voice, head and neck oncology, surgical and other ENT-related interventions for obstructive sleep apnoea, AI education and implementation, and AI-assisted patient education.
RESULTS: The searches identified 2,925 database records, four records from other sources and seven records from an updated search and expert suggestions; thirteen studies met the inclusion criteria. The evidence covered AI awareness and education, large language model (LLM)-based patient education, cochlear implantation, rhinology and skull base surgery, facial plastic surgery, laryngology and voice, and head and neck oncology; no eligible study addressed surgical or other ENT-related interventions for obstructive sleep apnoea. Clinicians were generally positive about AI, while clinical studies showed promising performance of AI models. LLM chatbots generally gave accurate postoperative advice, although, when validated tools were used, their responses were harder to read and less actionable than patient-education standards recommend. However, limited AI education, small or non-Saudi datasets, limited external validation, and implementation barriers remained important challenges.
CONCLUSION: AI research in Saudi otolaryngology remains limited but is expanding across several clinical domains. Future research should prioritise Saudi-specific datasets, multicentre clinical validation, structured AI education, rigorous bilingual (Arabic and English) evaluation of LLM-generated patient information, and safe integration of AI into clinical practice in line with Saudi Vision 2030.