Scholarlink

<p>Sleep staging plays a vital role in diagnosing sleep disorders, with obstructive sleep apnea (OSA) being one of the most prevalent conditions. Clinical data from OSA patients, including sleep duration, age, gender, BMI, and sleep position, has been found to significantly correlate with sleep stages. In this study, we proposed a large language model (LLM)-based clinical information analysis method utilizing OpenAI’s latest model, ChatGPT-4o, to analyze clinical data from 2570 OSA patients rigorously selected by medical experts. The results of this method indicate that sleep position is the most influential factor affecting sleep staging accuracy, accounting for 74.8%. We incorporated sleep position as supplementary input alongside single-channel EEG into our proposed model, SPSleepNet. The model was evaluated on OSA patients with different severity levels, including simple snoring, mild, moderate, and severe cases. Experimental results demonstrated that integrating sleep position information led to substantial improvements in macro-averaged F1-score, geometric mean, Cohen’s kappa, and accuracy. Additionally, high AUC values confirmed the robustness of our approach. This study highlights the significance of integrating clinically meaningful factors, especially sleep position, into sleep staging models. It provides a novel and effective pathway for improving sleep stage classification in OSA patients. Code is available at <a href="https://github.com/yykk187/spsleepnet">https://github.com/yykk187/spsleepnet</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

SPSleepNet: enhancing EEG-based sleep staging for OSA patients via sleep position integration

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月25日

查看全文
  • Yue Zhang,
  • Xihe Qiu,
  • Gengchen Ma,
  • Long Yang,
  • Jing Tao,
  • Jingjing Huang

摘要

Sleep staging plays a vital role in diagnosing sleep disorders, with obstructive sleep apnea (OSA) being one of the most prevalent conditions. Clinical data from OSA patients, including sleep duration, age, gender, BMI, and sleep position, has been found to significantly correlate with sleep stages. In this study, we proposed a large language model (LLM)-based clinical information analysis method utilizing OpenAI’s latest model, ChatGPT-4o, to analyze clinical data from 2570 OSA patients rigorously selected by medical experts. The results of this method indicate that sleep position is the most influential factor affecting sleep staging accuracy, accounting for 74.8%. We incorporated sleep position as supplementary input alongside single-channel EEG into our proposed model, SPSleepNet. The model was evaluated on OSA patients with different severity levels, including simple snoring, mild, moderate, and severe cases. Experimental results demonstrated that integrating sleep position information led to substantial improvements in macro-averaged F1-score, geometric mean, Cohen’s kappa, and accuracy. Additionally, high AUC values confirmed the robustness of our approach. This study highlights the significance of integrating clinically meaningful factors, especially sleep position, into sleep staging models. It provides a novel and effective pathway for improving sleep stage classification in OSA patients. Code is available at https://github.com/yykk187/spsleepnet.