基于指数加权的时变Hurst指数的睡眠脑电信号研究

Research on Self-similarity Time-varying Hurst Index Estimation Method Based on Exponential Weighted Method

睡眠脑电信号分析对于脑疾病早期诊断和睡眠质量监测具有重要意义。睡眠脑电信号具有明显的长相关特性,Hurst指数估计被广泛用于表征时间序列的分数或缩放特性、成为预测长相关时间序列的有效方法。但恒定的Hurst指数不能捕获动态睡眠EEG信号的详细信息,因此本文引入时变Hurst指数的指数加权方法,在移动时间窗口上对数据进行指数加权处理,对睡眠EEG信号时间序列进行动态分析。分析结果表明,基于指数加权的时变Hurst指数估计为不同睡眠阶段的大脑活动动态评估和脑疾病早期诊断提供了一种新颖而有效的研究方法。

Sleep EEG signal analysis is of great significance for early diagnosis of brain diseases and monitoring of sleep quality.Sleep EEG signals have obvious long correlation characteristics.Hurst index estimation is widely used to characterize the fractional or scaling characteristics of time series and become an effective method to predict long correlation time series.However,the constant Hurst index can not capture the detailed information of dynamic sleep EEG signal,so this paper introduces the exponential weighting method of time-varying Hurst index to process the data exponentially in the moving time window,and to dynamically analyze the time series of sleep EEG signal.The results show that time-varying Hurst index estimation based on exponential weighting provides a novel and effective method for dynamic evaluation of brain activity and early diagnosis of brain diseases at different sleep stages.