Scholarlink

<p>A well-designed sleep environment is essential for meeting occupants’ needs for high-quality sleep. With advances in artificial intelligence, intelligent regulation of indoor environment targeting sleep quality has become increasingly feasible. However, related research remains very limited. Therefore, based on three experimental studies involving healthy adults, this study developed an intelligent sleep environment control system by integrating sleep staging and environmental recommendation models. A bidirectional long short-term memory (BLSTM) neural network was employed to construct the sleep staging model, incorporating environmental and individual factors with electrophysiological signals through a sequence attention mechanism. A recommendation algorithm was further employed to extend the application of the model to sleep environment regulation. Results showed that the proposed model achieved an overall accuracy of 80.6% in predicting sleep stages at the next time point, with accuracies of 86.7% for deep sleep and 77.7% for rapid eye movement (REM) sleep. The ablation experiment further showed that the sequence attention mechanism improved the overall accuracy by 4.8%. Given its limited prediction performance for the Wake stage (40.2%), the current model is primarily applicable to the sleep maintenance period. Several electrophysiological signal features exhibited significant differences across environmental parameter ranges, and all of these differences demonstrated medium to large effect sizes. In addition, environmental recommendation results indicated that the environmental parameters corresponding to the longest hourly durations of slow wave sleep (SWS) and REM sleep varied throughout the night. Furthermore, gender differences in optimal sleep environments were observed in both halves of the night (first half: <i>p</i> = 0.037; second half: <i>p</i> = 0.044), along with seasonal differences in optimal sleep temperatures for males during the second half of the night (<i>p</i> = 0.034). Finally, an intelligent sleep environment regulation strategy was proposed, which provides a foundation for future precise and personalized sleep environment regulation.</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

A method for intelligent sleep environment regulation

  • 期刊论文

  • 发表日期：

    2026年7月10日

查看全文
  • Xudong Zhang,
  • Ting Cao,
  • Junmeng Lyu,
  • Zhiwei Lian

摘要

A well-designed sleep environment is essential for meeting occupants’ needs for high-quality sleep. With advances in artificial intelligence, intelligent regulation of indoor environment targeting sleep quality has become increasingly feasible. However, related research remains very limited. Therefore, based on three experimental studies involving healthy adults, this study developed an intelligent sleep environment control system by integrating sleep staging and environmental recommendation models. A bidirectional long short-term memory (BLSTM) neural network was employed to construct the sleep staging model, incorporating environmental and individual factors with electrophysiological signals through a sequence attention mechanism. A recommendation algorithm was further employed to extend the application of the model to sleep environment regulation. Results showed that the proposed model achieved an overall accuracy of 80.6% in predicting sleep stages at the next time point, with accuracies of 86.7% for deep sleep and 77.7% for rapid eye movement (REM) sleep. The ablation experiment further showed that the sequence attention mechanism improved the overall accuracy by 4.8%. Given its limited prediction performance for the Wake stage (40.2%), the current model is primarily applicable to the sleep maintenance period. Several electrophysiological signal features exhibited significant differences across environmental parameter ranges, and all of these differences demonstrated medium to large effect sizes. In addition, environmental recommendation results indicated that the environmental parameters corresponding to the longest hourly durations of slow wave sleep (SWS) and REM sleep varied throughout the night. Furthermore, gender differences in optimal sleep environments were observed in both halves of the night (first half: p = 0.037; second half: p = 0.044), along with seasonal differences in optimal sleep temperatures for males during the second half of the night (p = 0.034). Finally, an intelligent sleep environment regulation strategy was proposed, which provides a foundation for future precise and personalized sleep environment regulation.