本研究方案旨在对比大语言模型（如DeepSeek GPT）与专病GPT（如BAIMGPT）在阻塞性睡眠呼吸暂停（OSA）咨询和管理中的临床应用效果。采用多中心真实世界研究设计，纳入1000例OSA患者或高风险人群，通过用户横向评价和第三方专家审核，评估两种模型在便捷性、亲切感、安全感、问题理解准确性、回答准确性、语音交互、视觉赋能及需求程度等方面的表现。研究重点关注两种模型在OSA筛查、诊断准确性和个性化防治中的作用，并探讨其在提升患者教育、医生培训和基层医疗覆盖方面的潜力。研究结果将为人工智能技术在OSA 诊疗中的优化应用提供实证依据，推动精准医疗和健康管理的发展。

This research plan aims to compare the clinical application effects of large language models (such as DeepSeekGPT) and specialized disease GPT (such as BAIMGPT) in the consultation and management of obstructive sleep apnea (OSA). A multicenter real-world research design is adopted, involving 1000 OSA patients or high-risk individuals. Through user cross-sectional evaluations and third-party expert reviews, the performance of the two models in aspects such as convenience, friendliness, security, accuracy of problem understanding, accuracy of answers, voice interaction, visual empowerment, and the degree of patient needs is assessed. The research focuses on the roles of the two models in OSA screening, diagnostic accuracy, and personalized prevention and treatment, and explores their potential in enhancing patient education, doctor training, and coverage of primary medical care. The research results will provide empirical evidence for the optimized application of artificial intelligence technology in OSA diagnosis and

treatment, and promote the development of precision medicine and health management.