基于互信息和自相关函数的睡眠期间心率变异性研究

目的：研究互信息和自相关函数在睡眠各阶段心率变异性（heart rate variability，HRV）分析中的应用。方法：采用网络公开数据库SleepHeart Rateand Stroke VolumeDataBarhk，将RR序列分为30S一段，并以每30S数据为中心截取5minRR序列作为待分析对象。提取5minRR序列的互信息特征BDM、PDM和自相关函数特征BDC、PDC，然后用统计方法分析各特征在觉醒瓜EM／浅睡／深睡四种睡眠状态下的差异。结果：浅睡和深睡期间，BDM、PDM显著高于觉醒和REM睡眠（P〈0．001）。而BDC显著低于觉醒和REM睡眠（P〈0．001），PDC无显著差异（P〉0．05）。结论：①BDM、PDM和BDC从不同角度反映了不同睡眠阶段下RR序列的特征，它们或许与HRV不同的调节机制有关。@BDM、PDM和BDC可作为辅助HRV睡眠分期的新指标。

Research on the Heart Rate Variability during Sleep Based on Mutual Information and Autocorrelation Function

Objective： To research the application of mutual information and autocorrelation function on the analysis of heart rate variability （HRV） during different sleep stages. Methods： The online public database Sleep Heart Rate and Stroke Volume Data Bank was adopted. The whole record was segmented into 30 s epochs at first, and then 5 min segments center around each 30 s epoch were extracted as data to be analyzed. Afterwards, the mutual information features, i.e. BDM and PDM, and autoeorrelation function features, i.e. BDc and PDc of RR sequence were determined for each 5 min epochs. At last, the significance levels between different sleep stages （wake/RENFlight sleep/deep sleep） for every feature were evaluated through statistical method. Results： BDM and PDM during light sleep and deep sleep were significantly higher than those in wake and REM sleep （P〈0.001）, while BDc were significantly lower than those in wake and REM sleep （P〈0.001）. And no significant difference of PDc was showed among different sleep stages （P〉0.05）. Conclusion：① BDM, PDM and BDc could reflect the features of RR sequence during differem sleep stages, which might be related to different regulation mechanism ofHRV. ②BDM, PDM and BDc could be used as new indicators for sleep stages classification based on HRV.