Abstract: Sleep apnea seriously affects the quality of life and health. Polysomnography (PSG) is the "gold standard" for diagnosis, but it is expensive and inconvenient for long-term monitoring. This paper proposes a new method based on a universal smart wristband to detect sleep apnea. By analyzing the heart rate, blood oxygen saturation, and sleep state data collected by the wristband, an adaptive physiological data reconstruction method and a data interpolation method are used to achieve noise filtering. In feature engineering, continuous physiological variables and categorical variables are fused to deeply extract sleep state features. The classification module uses a lightweight gated recurrent unit model to simplify the training process and reduce the risk of overfitting. The experiment obtained 93.68% accuracy and 93.97% recall on a 23-person dataset, which is 13.56% higher than the baseline method. Correlation analysis find that blood oxygen saturation, body mass index, and age are confirmed as key features for determining sleep apnea. Compared with polysomnography, this method is more suitable for long-term monitoring in a home environment.

Key words: sleep apnea detection, universal wristband, multimodal data processing, long-term health monitoring, analysis of multi-factor impact indicators

摘要： 睡眠呼吸暂停严重影响生活质量和健康。多导睡眠图（PSG）是诊断的“金标准”，但成本高且不便长期监测。基于此，一种基于通用运动手环的新方法被提出以便捷化检测睡眠呼吸暂停。通过分析手环采集的心率、血氧饱和度和睡眠状态数据，采用自适应生理数据重构方法和数据插值方法，实现噪声滤除。在特征工程中，融合连续生理变量和类别变量，以深度提取睡眠状态特征。分类模块采用轻量级门控循环单元模型，简化训练过程，降低过拟合风险。实验在23人数据集上获得93.68%准确率和93.97%召回率，比基线方法高出13.56%。相关性分析发现血氧饱和度、身体质量指数和年龄被确认为是判断睡眠呼吸暂停的关键特征。与多导仪相比，该方法更适用于家庭环境下的长期监测。

关键词: 睡眠呼吸暂停检测, 通用手环, 多模态数据处理, 长时健康监测, 多因素影响指标分析